I grew up in a college town. My dad, like most of my friends’ parents, worked for Dartmouth College, by far the dominant employer in Hanover, N.H. Many of my friends were faculty brats, with one or both parents working as professors at Dartmouth. But my dad worked as a programmer.
Everything about programming today is better than it was in 1987. When Dartmouth hired him that year, most of his work involved writing software for machines like the Macintosh SE, which boasted a maximum of 4 megabytes of memory. Today, to run my dad’s software, you need Mac OS X 10.5, which requires 512 megabytes of memory. I run it on a machine with 6 gigabytes of memory - more than a thousand times as much as the Mac SE's maximum. The machines he works with now are hundreds, often thousands, of times more capable than the ones he worked with in 1987. The field is getting more productive.
By contrast, my faculty brat friends’ parents have seen their jobs change…not at all. In 1987, they walked into stuffy rooms and delivered soliloquies on their subject of expertise to seated undergraduates. In 2013, they do the exact same thing. In 1987, they researched, wrote and submitted articles to journals published twice or maybe four times a year. In 2013, they do the exact same thing.
When universities and their faculties try to explain how spending has gotten so out of control at research institutions, they generally point to this discrepancy. Other fields get more productive. College teaching doesn’t. This phenomenon is known as Baumol's cost disease.
Baumol’s disease sounds like a horrible and obscure malady, something that both cripples its victims and makes them go into lengthy explanations of it at dinner parties. It’s actually a horrible and obscure malady that occurs when one swath of the economy, for whatever reason, just can’t get more productive. And yes, economists tend to go into lengthy explanations of it at dinner parties.
“Productive” here is meant in the most technical of senses, as the amount of goods and services each worker can generate for his employer over a given period of time. Normally, when productivity increases, wages increase, too. That connection makes sense for both the employer and employee. If the worker is able to produce more value at his job, he demands more money. If the boss refuses to pay, the worker can get more money elsewhere. But if the worker isn’t getting more productive, then why would his boss agree to give him a raise?
Because he can go somewhere else. William Baumol, who is, amazingly, still an active economist at age 90 and just released a book on these very issues, realized in a series of papers in the 1960s (many coauthored with future Princeton President William Bowen) that industries that didn’t see productivity increases might still see costly wage increases. “The insight that Baumol and Bowen had was that there are some industries where productivity growth is very likely and some where it’s not,” Robert Archibald, an economist at the College of William & Mary, explains.
Performing a string quartet will always require two violinists, a violist and a cellist. You can’t magically produce the same piece with just two people. Higher education, for at least the past few millennia, has seemed to fall in this category as well. “What just happened in my classroom is not very different from what happened in Plato’s academy,” quips Archibald. We’ve gotten better at auditorium-building, perhaps, but lecturers generally haven’t gotten more productive.
But other fields do get more productive. We’ve become much better at, say, manufacturing huge quantities of chemicals than we were 3,000 years ago. And correspondingly, if you’re good at synthesizing large amounts of polyurethane, you’re going to get paid a lot more today than you were 30 or 40 years ago, because productivity is growing in that area.
What really drives the Baumol explanation, Archibald continues, is the fact that “the industries where productivity growth is likely, and the industries where productivity growth is not very likely, have to compete for workers.” A star doctoral candidate in chemistry could go work for DuPont on making more and better plastics, or he could do similar research at a university.
If the university is to have any hope of snagging workers like that, it’s going to raise its wages above where the industry’s productivity level would suggest it should place them. To pay for that, the university must increase its fees or other revenue streams. So you have costs growing, even as the university gets no more productive. That’s a recipe for ever-increasing costs with no noticeable return. This situation is what’s come to be known as Baumol’s cost disease.
It’s clear enough to see how this could work in the case of higher education. It’s a sector that needs lots of highly trained workers and has to compete with much more productive industries to get them. Archibald and his colleague David Feldman advance a sophisticated version of the Baumol hypothesis in their book, Why Does College Cost So Much?
Higher education, they note, is a service that is highly labor-intensive, and which requires skilled labor in particular. And in recent decades, the cost of skilled labor has shot upwards relative to the cost of unskilled labor; in other words, college graduates started to pull away from their high-school educated peers. So it should come as no surprise that colleges would see costs rise. The cost of their main input, people, was rising a lot. The cost of college, they continue, has increased at about the same rate as other skill-intensive services. Nothing nefarious is going on. It’s just Baumol’s disease working its dark magic.
It’s a popular theory. A solid plurality of the economists on IGM’s Economic Experts Panel endorsed it last year. But it just can’t explain the growth in tuition and spending in recent decades.
If Baumol’s disease is at work, you’d expect to see certain patterns in the data. The amount that universities are spending in professor salaries and compensation should be going up, for one thing. What’s more, that increase in compensation should be the only thing accounting for increases in university spending.
Do we see that? It’s a bit hard to say, because the data we have breaking down college costs is limited. The best source, the Delta Cost Project, however, can provide some clues. Delta breaks costs down into a number of categories, one of which is “instruction,” of which the major expense is faculty salaries and benefits. If Baumol’s disease holds, you should expect that to account for most if not all of the increase in per-student expenditures.
Does it? Not really. Instruction grows more slowly than educational costs as a whole. Take public research universities. Between 2000 and 2010, spending on operations and maintenance and financial aid fell by $806 per student, but other costs increased by $4,723, so costs rose overall. Of that $4,723 increase, 41.5 percent went to increased spending on "auxiliary enterprises" like hospitals, 22 percent went to increased research spending, 20.7 percent went to support and public service spending, and the rest went to increased costs of instruction. That just isn't what the picture should look like if Baumol's cost disease is the core problem.
Now, you could object here that many of these other categories require highly educated laborers. Administrative deans, after all, tend to have bachelor's degrees, hospitals are afflicted with the high growth of health care costs, and so forth. But note that that's different from the Baumol thesis, which is that the core act of teaching is leading to the inexorable increases in costs.
The glut of PhDs in recent years also pushes against the theory. Baumol cost increases should only occur when PhDs are in a position to demand a certain wage from the universities that employ them. But now that there are many more PhDs, especially in the humanities, than tenure-track spots for them, the salaries of instructors are actually falling as universities turn to adjuncts. That directly contradicts the Baumol explanation.
Robert Martin at Centre College has done some of the most compelling research on this, and in a paper published this year, coauthored by Carter Hill, showcased many other data points that should generate skepticism about the Baumol hypothesis. For example, between 1987 and 2008, full professor salaries increased by an average of 0.9 percent a year, and assistant professor salaries by a measly 0.7 percent a year. But support staff salaries grew at more than double that rate. If the core act of teaching is driving costs higher, that just shouldn't be happening.
As Archibald and Feldman write in their book, "An institution can increase class size to raise measured output (students taught per faculty year) or it can use an increasing number of less expensive adjunct teachers to deliver the service, but these examples of productivity gain are likely to be perceived as decreases in quality, both in the quality rankings and in the minds of the students."
The same is not true of changes in employment of service staff. It's implausible to argue that a school couldn't, say, reduce the pay of administrative deans it employs without also reducing educational quality. For Baumol to hold, teaching itself has to be what's driving costs up. That doesn't appear to be the case. “For most administrative roles, the argument for why you can’t have productivity increases breaks down,” Education Sector’s Andrew Gillen explains.
So what does explain the increase, if the Baumol hypothesis doesn’t? The best candidate, I think, is something called the Bowen hypothesis. We’ll talk about that tomorrow.