Universities often report a number that appears to indicate how much the university spends on instruction. We might believe that this number accurately represents teaching expenses and even do some analysis based on that belief. We would be wrong to do so.
This somewhat cynical observation by Lombardi was informed by his broad and sometimes painful experiences as Provost at Johns Hopkins University, President of the University of Florida, Chancellor of the University of Massachusetts at Amherst, and President of the Louisiana State University System. However, in these times of heated discussions over who should pay for higher education and a background of rapidly increasing student debt, it is important to have some idea of what the actual costs of producing that education are. In this post, I review some of the reasons why it is difficult to define the instructional costs at a research university, and why various constituencies might not want that information to be generally available. After discussing how a business model view simplifies some of the issues around calculating instructional costs, I describe a recent analysis of such costs in the University of California system, which reaches some surprising conclusions. These conclusions lead to a consideration of why cost -shifting between missions is so important in the current approach of the research university. Taken together, these results suggest that one of the key issues that should be focused on in order to control higher education prices are the synergies between the different functions of the research university and the actual "added value" to the customer of those synergies. In particular, the analysis suggests that rising prices in undergraduate education are not likely be controlled unless society finds alternative ways to fund a significant component of the cost of university research.
Readers of this blog know that Disrupting College by Clayton M. Christensen, Michael B. Horn, Louis Caldera and Louis Soares (CHCS) reveals some key aspects of the operations of a research university. In particular, CHCS show that the research university operates three different generic types of business models simultaneously as it carries out its missions of teaching, research, and social growth of students. This type of multiple business model operation leads to very high operational overhead rates, leading to high costs. In part this overhead is due to the fact that the constraints imposed by operating several business models simultaneously means that no one of the models can be optimized with respect to cost. It is important to recognize that these constraints also make it impossible to optimize any of the models with respect to quality of output.
Thus, one part of the difficulty of defining educational costs is the interlocking business models that lead to individuals and facilities being used in multiple activities. In addition, as I pointed out in A business model view of changing times in higher education :
...when running several business models simultaneously, there is great opportunity for cost shifting from one component to another when the organization wants to discreetly cross subsidize activities. It thus becomes very difficult, if not impossible, for customers to understand exactly what they are paying for.
That is, some of the difficulty in determining costs of individual components of the operation is that we prefer it that way - it lets us put our thumbs on the scale to reflect the way we in higher education prioritize the component activities of the university.
We enable (encourage?) this indeterminacy through our accepted form of higher education accounting - fund accounting. The focus of fund accounting is on management of individual sources of income, rather than on the expenses related to particular activities. As Lombardi points out;
Although fund accounting does not prevent universities from understanding their finances, it does not require them to do so.
Cost accounting, on the other hand, focuses on understanding the actual costs of a product, so that management can manage those costs. As such, it is primarily a management tool for the creation of efficiency and informed decision making. So, to find out the cost of education in a university, we have to change our mindset and do cost accounting.
It is important to recognize that this approach of intertwined costs has been very successful overall for America and its higher education system. Allowing university administrators to follow their own priorities in cost shifting and cost sharing between activities such as research and teaching has resulted in an enormously successful university research enterprise, and a highly respected educational system. In a system in which costs are transparent and cost shifting does not occur, the priorities and the balances will be determined by the collective priorities of the payers - students and parents, government, foundations, etc.Outcomes could be quiet different from those currently defined by the providers!
In a conservative "don't fool with success" mode, one could argue that we should not try to clarify the actual uses of resources. Two reasons why this is not the appropriate response at this time jump out. First, there is an enormous pressure on higher education to lower educational costs, which have grown much faster than inflation for many decades. The current system, which demands large annual real cost increases, is broken. It is very difficult to do experiments to lower costs of education if administrators do not know the original cost and the contributions of the various components of education to that cost. Does a new approach lower costs or increase costs? Does a change of emphasis or priority raise or lower cost? For this, cost accounting with its transparency is absolutely essential. Second, with the increasing emphasis on having students (and or their parents) pay for their higher education, they have every right to demand to know what the actual cost of producing that education is. They should not be going into debt to fund other activities of the university unless they choose to do so.
Business Model Insights for Cost Accounting in
Of course, it is not trivial to isolate costs when there are multiple business models running simultaneously. In the university case, faculty perform multiple functions, as do facilities such as classrooms, labs, and libraries. Some have argued that the research university is a classic example of the joint products problem, in which multiple outputs are created from a single input using the same general process. In such cases, assigning costs up to the divergence into different outputs is essentially arbitrary and can be shown to offer almost no management value. However, I argue that this is not a joint products problem, and that the business model approach of CHCS suggests a different perspective: this perspective facilitates calculation of specific costs by sharpening the definition of what is to be calculated, and, more important, highlights some of the key issues that call for increased transparency and discussion.
A business model describes how (processes and resources used) and why (who the customers are and what they want they want the product to do) a product is made and its cost of production. So a logical use of cost accounting in a multiple business model organization is to analyze the different business models present in the university one by one. (For a detailed view of the business model approach to higher education, see A business model view of changing times in higher education.)
In this approach, we begin by considering the business model for the educational function , and then focus on how that model interacts with the research business model. We will pay little attention to the student social growth business model, which is of importance itself because it is the origin of much of the "arms race" that helps to drive up college costs. However, it does not raise many of the critical issues presented by the other two models.
Education, according to CHCS, follows the generic Value Added Process business model, in which (generically speaking) an incomplete input is transformed into an output which is more complete and of higher value. For the traditional student body, there are really two broad but distinct inputs to this model: undergraduates, and graduate and professional students (in fact, a thorough cost accounting could break those those two inputs into finer categories). These two distinct "inputs" are associated with different value propositions, and move through the system following somewhat different processes. In fact, they follow different business models (both of the same generic Value Added type). Consequently, assigning costs for the components of the educational function itself is not an arbitrary joint products exercise without management utility, but is a definable exercise that can have significant value for the types of management decisions needed to control spiraling tuition.
Perhaps the most important consequence of doing a cost analysis based on the various business models is that it shows that the discussion should not be focused on determining the cost of the individual business models. Instead, the critical discussion really needs to focus on the value added by the presence of multiple business models - and for whom the value is added.
Cost Accounting for the University of California System
An example of a cost accounting approach is given in two interesting articles by Charles Schwartz, Professor Emeritus at the University of California (UC) Berkeley. In these articles, Schwartz attempts to quantify actual undergraduate educational costs in the UC system: Cost Accounting for the Academic Missions at the University of California, and Reform the Funding Model for the University of California. He bases his calculations on analyses of numerous public budget documents of the UC. I will say that most published budget documents of all kinds are meant to give the minimal required information while artfully obscuring information that leadership would prefer not be in the public domain: reading of several of the UC documents used by Schwartz shows that the UC truly is a master of this approach. Consequently, I can't speak to the accuracy of Schwartz's derived data (he has spent years trying to decipher these documents, I have not). However it is obvious that his approach is appropriate (with one minor caveat from the perspective of this post) and serves to point out some of the very important questions about the interplay of the different business models that rightfully should be addressed publicly.
Schwartz's approach is to build the undergraduate cost budget from "the ground up" by separating undergraduate related expenses out the general UC budget. Obviously, a large expense item in the budget of a research university is the faculty salary line. Schwartz begins by using an existing UC faculty survey of workload to break down average faculty salary into components of undergraduate teaching, graduate teaching, research, and service. Not unexpectedly, the result is close to the "generic" research university assumption of 50% teaching - divided between undergraduate and graduate - and 50% research. In the UC case, faculty reported that the teaching time was roughly evenly split between undergraduates and graduates. Thus about 25% of faculty academic year salary should be allocated to the cost of undergraduate education and a similar amount to the cost of graduate education in any cost analysis. He also divides up expenditures of adjunct faculty and TA's along the lines of course assignments to decide how to allocate that component of the budget. He makes guesses as to the appropriate ways to allocate costs of libraries, student services, etc. among the functions. (I suspect that quiet good estimates of these allocations have been obtained for the UC for indirect cost negotiations, but those numbers are not publicly available.) In this last step, his calculation differs somewhat from what I suggest above in that e.g. many student services costs are actually part of the student growth business model. However, that makes little difference for the purpose of this discussion.
To make the next very important step, I must digress a bit and talk about the various types of research. There basically are three categories of research that pop up in accounting. The two well defined categories both belong to Organized Research: Sponsored Research and University Research. As defined in OMB Circular A21:
(1) Sponsored research means all research and development activities that are sponsored by Federal and non-Federal agencies and organizations......
2) University research means all research and development activities that are separately budgeted and accounted for by the institution under an internal application of institutional funds......
That is, both of these types of research appear as specific defined projects in a budget that are accounted for at the end of the budget cycle; university research funds are typically dispersed through an internal competitive application and award process. The third type of research expenditure, called Departmental Research, is "everything else". Into this category go such university expenditures as start up funds for new faculty, seed funds for new research projects, bridge funds between grants, funds for visiting researchers, course release for faculty so that they can do their own research, and faculty specific gift accounts. Thus, Departmental Research can represent a very substantial faculty research-related expenditure for a major university.
Unfortunately, A21, which focuses on the organized research of an institution, contains the phrase
Departmental research, for purposes of this document, is not considered as a major function, but as a part of the instruction function of the institution.
This phrase echos a position taken in an earlier report by the National Association of College and University Business Officers (NACUBO), Explaining College Costs. This report was written to provide guidance to colleges and universities in explaining why tuition is so high: costs of education are even higher. To make this argument, it was desirable to have a uniform approach to calculating educational costs so that most institutions could say similar things. One might cynically argue that it was critical to the argument to define terms so as to make the calculated cost as high as possible. In any case, the approach chosen was to include all of Departmental Research in the cost of education. This brings up some major questions to be addressed later, but suffice it to say, thanks to A21 and NACUBO, in most university budgets Departmental Research is absorbed into the Education category and can no longer be broken out easily.
Despite this obstacle, Schwartz makes an attempt to remove Departmental Research from the calculation of undergraduate teaching costs. Putting all of his numbers together, he concludes that the actual amount of spending per undergraduate student on the education function at the UC in 2013-2104 is $7,500. This number is considerably less than the 2013-2014 average expenditure per student (averaged over both graduate and undergraduate students) of $18,060 reported by the UC. Schwartz's calculated expenditures for undergraduate education are, in fact, even less than the tuition and fees currently required of the students! As far as I can tell, neither Schwartz nor the UC include major capital costs (e.g. buildings and their replacements) in their calculations, probably because that comes out of another budget in the UC system.
Although Schwartz takes into account a large number of effects, it is revealing to note that most of the difference between his calculated costs and those of the UC apparently relate to only two effects: his attribution of only 1/4 of faculty academic year salary to the cost of undergraduate education, and excluding Departmental Research from the calculation. Those are clearly appropriate decisions for a cost accounting of undergraduate teaching business model. In a similar vein, 1/4 of faculty academic year salary should be attributed to graduate teaching in cost accounting for that business model, and 1/2 of faculty academic year salaries and Departmental Research should be combined with Organized Research in the cost accounting of the research business model. That is, Schwartz's calculation suggests that the UC calculation has shifted costs equal to about $10,000 per undergraduate from the graduate education and research functions to the undergraduate instruction function - or about $2B overall for the UC.
Schwartz's calculation suggests that doing a cost accounting analysis of the major business models represented in a research university is conceptually relatively straightforward and has the benefit of clarifying some of the aspects of cost sharing and cost shifting that can occur in a multi-business model budget.
The Importance of Cost-Shifting in Maintaining the Status Quo
of the Research University
I mentioned above that the cost-shifting from research to teaching was in the overall interest of the research university. The reason that this is true is that reputation- building research is very expensive - in fact, more expensive than society has been prepared to pay directly thus far.
We saw above that the funders of research do not pay for the 50% of faculty academic year salary that supports research in a research university, nor the very important components of Departmental Research such as new faculty start-up funds. In addition, a major university expenditure is the University Research component of Organized Research. University Research includes, as noted above, internal funding of research projects that are competitively awarded and evaluated at their end, and also cost sharing expenditures which may have been necessary to obtain some sponsored projects, and a very nasty thing called incomplete recovery of indirect costs.
This last item is worth discussing because it brings up a few misconceptions about sponsored research - e.g. that it is revenue neutral (or even positive) for the institution. Direct costs of organized research are those easily attributed to a specific project, such as equipment, materials, summer salary for faculty, postdoc salaries, and travel. However, there are additional costs of the research that are harder to attribute to a specific project, such as building maintenance, support services such as libraries, and general university administration - the indirect costs. These costs are attributed to specific projects through use of an average indirect cost rate that is obtained by dividing the total university indirect costs for organized research by the total direct cost of organized research. Each grant is then composed of a direct cost component and an indirect component that is obtained by multiplying the direct component by the indirect cost rate. This procedure in principle then leads to each university being able to achieve "full recovery" of its total indirect and direct costs for organized research when summed over all organized research projects - that is, the organized research will be revenue neutral for the university. Of course, the University Research component of the organized research has to absorb its proportional share of the total indirect costs, as should every organization that supports some component of the organized research.
Unfortunately, not every organization that provides research support is willing to pay full indirect costs appropriate to the direct cost funds that they provide. Most foundations and corporations, for example, will not, and state and local governments often will not either. This "incomplete recovery of indirect costs" leaves some portion of the actual total indirect expenditures for organized research uncovered by external funds - expenditures that therefore must be covered by internal university funds. That is, organized research is revenue negative, not revenue neutral.
The NSF report on expenditures for R&D in FY 2012 provides details that enables us to see the magnitude of these various university research expenditures summed over all research universities. University internal expenditures for R&D that year totaled about $13.6B; for comparison the total Higher Education Federal R& D for that year was about $40B. That is, universities' internal research support was almost 35% as large as the total university research support provided by the federal government. If we assume most of this research occurred in 200 institutions, then the average research university put $70M of its own funds into research of its faculty! The total university research expenditures of $13.6B are divided into the categories:
institutionally financed research $7.7B;
cost sharing on sponsored projects $1.3B;
incomplete indirect cost recovery $4.6B.
"Institutionally financed research" is defined in this NSF report as including competitively awarded internal grants, startup packages and bridge funding, other department funds, and tuition assistance for research personnel. No breakdowns are given of the contributions of these categories, but they reflect components of both University and Departmental Research as defined by A21.
Thus we see that a very large proportion of research costs in universities is not paid for by external entities, but must be covered by the universities themselves. The dollar amount of this university funded research is, in fact, growing rapidly, as shown for the period 2010-2013 in a Feb. 2015 NSF report.
Earlier NSF data show similar growth over the past several decades. Although some relatively small portion of these internal research costs may be covered by restricted endowments and the like, most of these costs must be paid out of unrestricted funds. Importantly, for most universities, by far the largest source of unrestricted income is student tuition or state funds nominally labeled as educational funds. As a consequence, the ability to do cost shifting between the research and education modules is very important to the health of the research university.
These data strongly suggest that there may be a causative relationship between rapidly rising undergraduate tuition and the research university's drive to do ever-more research.
Synergies of the Multiple Business Models in the
However, underlying functional cost does not necessarily represent the value that one of these products provides to a customer. Does the combination of research and undergraduate teaching in one organization provide benefits to the students that they should be willing to pay for? It is clear that the research university could not maintain its current organization if tuition were not a key supporter of the half time that faculty are paid to do research during the semester, and elements of Departmental Research such as faculty start up funds, course relief, bridge funding for research, etc.Without that support, the research university could not attract and keep the research active faculty that define the university. The question that needs to be addressed, then, "is do the undergraduate students really get that much benefit from the research programs that they are supporting?"
NACUBO strongly believes that they do. The report in which this cost shifting was endorsed argued:
Departmental research is vital and has a direct impact on the value and quality of instruction provided to students. Any arbitrary attempt to distinguish between departmental research and instruction ignores the fact that the integration of research and education is a major strength of the nation’s colleges and universities and directly benefits undergraduates.
Powerful words indeed, but unfortunately the benefit to undergraduates is not a "fact". I know of no data that supports the position that the quality of instruction provided by an active, front line researcher is necessarily better than that provided by a highly educated, but not research active professional - and some research that suggests that the opposite is true. In fact a significant body of research shows that the effectiveness of student learning overwhelmingly is a function of the pedagogical approaches used, whatever the research strengths of the instructor. In a research university emphasis is on research rather than teaching, and quality of teaching plays almost no significant role in tenure and salary decisions. Thus, there is little professional benefit to a faculty to improve his or her teaching by learning and practicing more effective new pedagogical approaches (although some faculty will certainly do so for personal satisfaction). This research focused reward structure suggests that it is certainly possible that the "direct impact" of departmental research on "value and quality of instruction provided to students" at a research university is in fact negative in general. This biased reward system and its outcomes are typical of the type of suboptimization of individual business models that occur when there are constraints provided by following several business models simultaneously.
On the other hand, when there are actual undergraduate research activities embedded in the education e.g. a senior research thesis, the research active faculty play a key role. Unfortunately, most undergraduates even at research universities do not participate significantly in such research intensive activities. No research university has embraced truly embedding research into the undergraduate curriculum as has The College of New Jersey, which has a very different faculty reward structure.
It may well be that there is a link between faculty research and some desired outcome of undergraduate education, but it has not yet been demonstrated by data, and so is not a "fact" but rather a "hopeful hypothesis" which happens to support the overall interests of the research university itself.
However, there certainly are indirect ways in which faculty research may provide value to undergraduate students. The reputation of a research university is based primarily on the research visibility of its faculty (Reputation and brand in the changing world of higher education). A diploma from an institution with a strong reputation may well be of value in finding jobs or gaining entrance to graduate schools. In addition, the networks that research-visible faculty have can be important in facilitating entry of their students into top ranked graduate programs.
It is this component of the analysis - determining the value to the customers of the multiple missions of the research university - that is difficult and ambiguous, not the determination of the cost of the education itself. This central question has traditionally been obscured by a conflation of the two issues, with the result that the customer does not know what she and/or the state is paying for. This ignorance enables the university to follow a faculty reward structure that does not give high weight to making this synergistic value real.
Conclusion: Cost and Price
In a sustainable, well run non-profit enterprise, income = expenses summed over all activities. But this obviously need not say that price=cost for each product the enterprise produces. Of course, price of any product will ultimately be set by an interplay of cost of production and what the customer is willing or able to pay. Purchasers of the education product of research universities - students, parents, state and federal governments - are increasingly insistently saying that the price is already too high and that high annual price increases are rapidly making it worse. This strongly suggests that it is time to reconsider the pricing of the educational product and its relationship to the underlying costs.
As shown above, in the research university there is a large component of the price of education that is added based on the producer's rational that the research of the university adds value not captured in a straight cost of production calculation. This position is supported by the desire of the research universities to set the external price of their research output lower than the actual cost of the research.
CHCC point out that rapidly increasing prices are due to sustaining innovation without growth in sales volume. For the research university, most sustaining innovation is in the escalation of the arms race of student amenities (the student social growth business model) and in increases in research quality and volume (the research business model). The student amenities race is an issue for all of higher education, and is a subject for another discussion. However, this current discussion has suggested that growth in research also drives growth in educational price.
Thus, to address successfully either high tuition or the rapid growth in tuition, research universities will have to revisit the degree of inclusion of faculty research costs in the educational price. The first step in doing this should be to do a clear cost analysis that does not seek a priori to conflate the cost of education and the benefits to students of research. Then the focus can be on demonstrating those research related benefits to the customers of education and determining what supplement the customers are willing to pay for those benefits. It could well be that this discussion will suggest to some institutions that weightings of faculty reward structures need to be changed in order to actually produce the desired synergistic benefits between research and education.
At the same time, research universities need to help the buyers of research and society generally understand that buyers of education can no longer be asked to pick up significant unfunded costs of research. If research is valuable, then those receiving value should pay their appropriate share of the cost - including students who should not be asked to pay for more than the benefit that they actually receive.