Tuesday, July 19, 2016

Opposable Thumbs

A longtime reader wrote this week with a great question:

Between the subsidies on the one hand and the prestige pricing on the other, it's hard from the outside to figure out what it actually costs to deliver a college education. What's your rule-of-thumb figure?

I like this one a lot, even though I have no intention of answering it directly.  In fact, until some other questions get addressed, I’d oppose any rule of thumb.

First, be careful to make a distinction between “price” and “cost.”  (The question refers to both.)  I’ll use “price” to refer to what students and/or families pay.  “Cost” is the cost of provision by the institution.  If the two matched perfectly, the college would break even.  But that’s not how this industry works.  Third-party scholarships and Federal grants go directly to price, but have very little direct effect on cost.  Some people like to claim that the availability of scholarships and grants feed an arms race that feeds cost in turn; from what I’ve seen, there’s some evidence for that in the four-year sector, but almost none in the two-year sector.  

Public colleges charge less than cost, making up the difference through operating support from states and/or localities.  Well-endowed private colleges do the same, making up the difference through endowment income.  Poorly-endowed private colleges try to break even.  For-profits charge more than cost.  That allows for-profits to expand much more quickly than nonprofits, but it also makes enrollment drops hurt them more.  To be fair, for-profits are also taxable, which is a cost to which nonprofits are immune.  If large public universities paid property taxes, their balance sheets would look very different.

Next, define “a college education.”  That can mean a lot of things.  It’s a lot cheaper to educate a history or poli sci major than a nursing or automotive tech major.  (Budget hawks who suggest “vocational schools” as alternatives to college almost always get this wrong.)  A college with a high adjunct percentage can probably deliver classes less expensively than one with a lot of full-timers.  A college with lots of intercollegiate sports teams will typically spend more than a college that doesn’t.  Study abroad costs more.  Boutique or specialized programs cost more.  Location can matter, both directly -- housing costs for students -- and indirectly, in the salaries that have to be paid to keep people.  The presence or absence of unions will affect the bottom line, as will the 800 pound gorilla of higher education budgets: health insurance.  

We even need to define “student.”  A selective four-year residential college may have all or nearly all full-time students, so headcount and FTE will be pretty much the same.  Most community colleges have significant numbers of part-time students, so headcount and FTE will diverge.  Neither is a perfect measure.  If you look only at headcount to do a per-student figure, community colleges will look weird.  If you look only at FTE, though, you’ll understate some of the back-office costs that community colleges incur.  The truth is somewhere in between.

Hearkening back to Econ 101, we also have to distinguish between fixed and marginal costs.  If a college with a full campus and administration only has one student, that one student has to cover a whole lot of salaries.  Additional students allow the amortization of those costs over more tuitions.  Most community colleges have a core of full-time faculty who cost more than adjuncts, so the variable cost of instruction doesn’t vary linearly.  It’s high at the outset, then cheaper as enrollments grow beyond a certain point.  That makes enrollment drops uniquely painful.  (I’ve outlined the mechanisms here.)

I’m also assuming that we’re only looking at operating costs.  If you factor in capital costs -- which any new operation would have to do -- they’d be considerably higher.  

If you’re looking for what I sometimes call a “big, dumb number,” I’d just pick a few colleges in a given sector and divide their operating budgets by their enrollments.  That’s a hugely imperfect measure in any number of ways, but it’s reality-based.  It’s open to all sorts of objections, ranging from the definitional ones outlined above to a more philosophical objection that it normalizes the status quo.  (Hegel famously claimed that “the real is rational and the rational real.”  I respectfully disagree; the real can be terribly, persistently irrational.)  I would argue, for instance, that most community colleges are underfunded as they currently stand.  That underfunding manifests itself in a higher than optimal reliance on adjunct faculty, thin full-time staffing at every level, and a host of small compromises that vary from place to place.  

It also assumes that differences by sector are written into nature, which they aren’t.  As I’ve mentioned recently, the same Americans who argue that inequities across K-12 are objectionable take inequities across higher ed as normal and natural.  I’ve never even seen a principled argument for them; they’re just sort of assumed.  Costs follow from structures.  

Much of the debate around student debt gets these basic points wrong.  It assumes that higher debt leads to higher default, which is exactly backwards; balances under $5k are the likeliest to default, since they mostly represent dropouts.  It further assumes that colleges follow the economic logic of for-profit businesses.  Anyone who has worked at a community college for any length of time can speak to the centrality of “mission” in staffing and budgetary calculations.  And strikingly from the perspective of someone concerned about too heavy a reliance on adjuncts, the debate tends to assume that costs are a function of a lack of fiscal discipline.  That may be true in some places, but it doesn’t explain the ubiquity of adjuncts.  Reliance on part-time labor is a symptom of a much larger issue.

So no, I won’t take the bait and pluck a number out of the sky.  I’d oppose a rule of thumb like that.  But I’d ask anyone who would to first offer answers to the questions above.