Measuring Baumol and Bowen Effects in Public Research Universities

by on November 1, 2012 at 5:20 pm in Economics, Education | Permalink

That is a new paper by Robert E. Martin and R. Carter Hill:

We estimate three models of cost per student using data from Carnegie I and II public research universities. There are 841 usable observations covering the period from 1987 to 2008. We find that staffing ratios are individually and collectively significant in each model. Further, we find evidence that shared governance lowers cost and that the optimal staffing ratio is approximately three tenure track faculty members for every one full time administrator. Costs are higher if the ratio is higher or lower than three to one. As of 2008 the number of full time administrators is almost double the number of tenure track faculty. Using the differential method and the coefficients estimated in the three models, we deconstruct the real cost changes per student between 1987 and 2008 into Baumol and Bowen effects. This analysis reveals that for every $1 in Baumol cost effects there are over $2 in Bowen cost effects. Taken together, these results suggest two thirds of the real cost changes between 1987 and 2008 are due to weak shared governance and serious agency problems among administrators and boards.

For the pointer I thank Michael Tamada (who does not necessarily endorse the argument).

John Thacker November 1, 2012 at 5:31 pm

Those are some pretty dramatically high costs. So they suggest that two to three times as much of the real cost increases are due to internal factors (Bowen) rather than external macroeconomic effects.

This result, if I understand it, argues that we should be able to reduce the cost of college quite significantly through reforms, whereas a primarily Baumol cost effect result would argue that it would be very difficult to do so.

Anora November 18, 2012 at 1:36 am

Overall Jim had a good year. But what will he do in an off season where we don’t even know who will own the team and when it will acltauly get sold. I hope TribCo will let the check book remain open while the FCC decides if the Trib can be sold to Snell and that Jim continues to work his magic.

John Thacker November 1, 2012 at 5:34 pm

They notice a decline in clerical staff (which is perhaps understandable) but “also persistent declines in technical, skilled, and service/maintenance staff ratios.” However, the number of managers and professional non-academic staff increased, as did their salaries. Meanwhile, there was a great increase in part-time and contract teaching staff.

Bill November 1, 2012 at 7:32 pm

At around their early 50′s, tenured faculty who do not want to do the young man’s work of research look longlingly at higher paying administrative positions while retaining their tenured status.

Tenured faculty can be just as bad as tenured administrators, and are often the same folks.

dbeach November 1, 2012 at 7:36 pm

Isn’t that perfectly consistent with “weak shared governance and serious agency problems among administrators and boards”? They can want those unproductive administrative jobs all they want, but they’ll only be available to them if there’s some kind of agency problem.

Bob Knaus November 1, 2012 at 8:05 pm

WTF? I used to do management consulting for public sector organizations. Our specialty was analyzing IT departments. We typically saw a “span of control” ratio in the 9:1 to 13:1 range for the larger ones. Meaning, there was one administrator for every 9 to 13 line staff.

Now perhaps tenure track facutly members require more cat herding effort than programmer/network admin types. But 3 to 4 times as much? My mind is boggled.

DocMerlin November 1, 2012 at 9:08 pm

Its not about cat herding… “administrators” are there to generate revenue from the customers of the university (government and parents). They have very little actual control, instead what they do is shuffle paperwork and take care of compliance.

Gwyan Rhabyt November 2, 2012 at 1:12 am

At Research I and II universities there are also many graduate students, who both do work (conduct research and teach classes) and need to be taught (like a new employee who is always in training sessions). So the true ratio of administrators to academic workers is more in line with your span of control.

mkt November 2, 2012 at 3:13 am

Careful. Due to the limitations of the data (from IPEDS, a Federal system of surveys of colleges and universities), Martin and Hill had to define “administrators” as employees who were categorized as either “executive, administrative, or managerial” as well as “other professional”. There are a ton of people in that “other professional” category who are indeed professional and administrative, but who have virtually no supervisorial role. I’ve spent over 20 years as a college administrator, and got my first permanent supervisee about two weeks ago. But in Martin and Hill’s analysis, I would count as an administrator.

This does not mean that Martin and Hill were wrong to use that measure. They tried looking only at the exec/admin/mgr categories of employees, and quickly found what every institutional researcher knows: colleges use such a widely varying array of definitions for that category that there is very little comparability. One college might report 40 employees in that category and another very similar college will report 116.

The same problem of course could afflict the “other professional” category, but both theory and empiricism strongly suggest that there’s great consistency with what sort of employees go into that category. Computer programmers, yes. Data entry personnel, probably no unless their positions require training or education which moves them into the “professional” category. Systems analysts (is that term still used?) probably yes, unless the college is one that likes to put a ton of people into the exec/admin/mgr category.

Martin and Hill might have better called their category “professionals” rather than “administrators”. But they are correct in combining those two IPEDS categories rather than just looking at exec/admin/mgr.

mkt November 2, 2012 at 3:15 am

*greater consistency* is what I meant to say, not “great consistency”.

Neil Strickland November 2, 2012 at 3:23 am

Does anyone have a link to a more recent and detailed account of Bowen’s argument? The paper by Martin and Hill refers to Bowen’s book from 1980, which is available from Amazon but you cannot look inside. In any case, many things have surely changed since then.

mkt November 2, 2012 at 3:46 am

The details are rather boring and un-illuminating IMO. “Bowen’s Rule,” paraphrased, is this: universities try to get as much revenue as they can. And then they spend it all.

(And that’s why they cost so much.)

Obviously there are a lot of details left out of that, but except for the one about colleges also wanting to increase their endowments (so in fact, the profit-maximizing model might not be that far off, as opposed to the model where they maximize revenues and expenditures), I don’t think much has changed in higher education in the intervening decades, at least not at the more selective and better endowed colleges. I don’t know how accurate Bowen’s Rule truly is, but it’s probably about as accurate now as it was then. Yes we have seen the rise of for-profit colleges, so there has been entry into the market for higher education — but admission to the selective colleges and universities still proceeds along lines similar to those of the 1980s. And yes now we have MOOCs, MRU, EDx, Udacity, Coursera, etc. but so far they’ve had as much impact on admission into the most selective universities as the University as Phoenix has. That might be changing even as we speak, but although Tyler and Alex are betting on MRU (or at least are hedging their bets by not going all in on GMU), I remain a skeptic about how much these web-based classes will transform higher education.

mkt November 2, 2012 at 3:34 am

Martin and Hill have made an interesting contribution to the study of costs in higher education, but with large enough flaws that I do not think we can believe their conclusions.

They do a good job by relaxing the assumption of cost minimization, and they extend that insight by examining two different (but not contradictory) theories of cost increases, “Baumol’s cost disease” and “Bowen’s rule”.

But they pretty much stop there, and pre-emptively assume that those are the only reasons for rising costs in higher education (aside from the obvious one of rising factor costs). The increased number of “administrators” (who would be better labelled as “professionals”) may reflect more than simple cost-padding; they also represent more IT personnel, more campus safety officers, more mental health counselors, more food made to order in the student cafeteria (if you attended college before say 1990 and haven’t eaten in a college cafeteria since then, most of them have waaay better food than before).

They similarly make the narrow assumption that the higher costs represent economic rent that goes to either faculty or “administrators”. And thus do not allow for the possibility — really the fact — that at least some economic rent is going to the students. How do we know this? Econ 101. Many colleges — in particular the selective public colleges — have a price which is below the market clearing level. And all selective colleges, public and private alike, are by definition characterized by excess demand: they routinely turn away consumers by denying them admission. Bingo, economic rent. (Compare how happy your two neighbors are, the one with the daughter who got admitted into Harvard versus the one whose daughter got denied.)

There could indeed be cost-padding going on, a la Bowen’s rule. But Martin and Hill assume that it’s all economic rent, with the only question being whether the rent goes to administrators or faculty. They refuse to consider the possibility that some of those costs represent schools competing via quality to attract students, and the fact that some of the rent goes to students rather than the college.

John Thacker November 2, 2012 at 10:22 am

They refuse to consider the possibility that some of those costs represent schools competing via quality to attract students,

Well it is odd that the schools by this logic would be essentially over-investing in quality. As you note, they don’t necessarily need to attract more students. In any case, I agree that there’s some bundling going on here.

and the fact that some of the rent goes to students rather than the college.

This is certainly true. Some of the rent goes to students and is extracted from their parents or from the government, allowing them to live at a higher standard of living (really nice gyms, better food, etc.) under the excuse of getting an education.

Robert Martin November 2, 2012 at 12:13 pm

Carter and I appreciate your comments; they will be useful to the rest of our work.

Several noticed we did not discuss in detail the other two cost drivers: government mandates and bundling. That is an oversight; we should have at least acknowledged those in a footnote. We do mention them in the conclusions. Our primary goal was to 1) develop a methodology for deconstructing real cost changes over a given time frame based on the model’s control variables and 2) generate empirical evidence regarding the relative size of Baumol and Bowen effects. The optimal tenure track/administrator (actually as suggested, it should be professional staff) ratio was a fortuitous by product of the investigation. It showed up in every model specification from the original OLS runs to the fixed effects and random effects models. Since it would not go away, we explored the question more deeply. I would call your attention to the elasticity of cost with respect to the ttad variable: whenever the ratio varies substantially (either way) from the optimal the impact on cost is much higher. A final comment on the 3:1 optimal value: I think that value is high because shared governance over resource allocation is currently so weak; with a new governance contract that made the faculty’s role in resource allocation more explicit, the optimal ratio would be lower.

In addition, we did not discuss Bowen’s rule in great detail. For those seeking a complete discussion of the modern theory underlying Bowen’s Rule, please refer to my 2011 book titled The College Cost Disease: Higher Cost and Lower Quality. The foundations for Bowen’s Rule are the peculiar economics of experience goods (reputation’s rule and quality is associated with cost, when consumers are uncertain about quality), agency theory, and the break-even model. Further, there is a discussion of bundling theory in the book.

The comments about rents flowing to students vs rents to insiders are interesting. Consider the economics of signaling and exclusivity when quality is uncertain. By analogy, consider an exclusive night club, a club where the bouncers sort through the clientele based on how attractive they are. The bouncers are “casting” each night’s performance to make the experience as attractive as possible in order to make the waiting list of new clients in the future as long as possible. Being admitted is a quality signal in the market for sex and the admitted client does receive rents created by being admitted. That in no way diminishes the rents earned by the owners of the night club, since the purpose of restricting admittance is to generate higher rents. If it did not generate higher rents, the clubs would not do it.

Comments on this entry are closed.

Previous post:

Next post: