"So Nick, how come a free-market economist like you is acting like a central planner?"
I heard that a lot when I was an associate dean. Sometimes it was said to tease me. Sometimes out of genuine annoyance.
I had my reply ready.
"There are three ways to allocate resources. The market is best; soviet central planning is second best; and the Hobbesian State of Nature is by far the worst. And right now I'm trying to drag you miserable lot out of the Hobbesian State of Nature into soviet central planning. Maybe later I will try to get something like a market going".
The story out of Texas A&M has been doing the rounds of the blogosphere. Most seem to be against even reporting data on revenues and expenses of individual professors. God forbid actually using that data for anything.
The sarcastic response is too tempting: "You economics professors are all in favour of prices and markets for everybody else. But it would never work for economics professors, of course. Right."
But what I really want to do is switch roles and play a part I never get to play. It's now my turn to say: "You pointy-headed professors sitting in your offices don't have a clue about how resources actually get allocated in the real world of the ivory tower. I actually know something about it, because unlike you I've got the practical experience of actually doing it. You useless theoreticians don't have a clue. I see the big picture and you don't. So shut up and listen while I tell you what really happens."
That's what I'm going to do here.
Let's start with student/professor ratios. Whenever a professor (or student) opens his mouth to complain about student/prof ratios, it's almost guaranteed he will reveal he doesn't have a clue how student/prof ratios are determined. This is the first thing you need to understand about universities. Let me show you my little model.
Universities sometimes run a surplus or a deficit, but won't or can't do this forever. So let's cut to the chase, and assume total revenue = total expenditures.
Nearly all revenues come from teaching students, and are proportional to the number of students taught. It's basically a horizontal Average Revenue curve.(In my university, about 50% comes from student fees, and the other 50% comes from a government grant per student, which is effectively a voucher.) So total revenue = revenue per student x number of students.
Profs are the biggest expense. If you like, add in support staff, and classrooms and stuff, into some overall measure of teaching resources, and call it "profs". Total expenditure = expenditure per prof x number of profs.
After some very complicated mathematical economics, available on request, we can derive the equilibrium to this model as:
Students per prof = (expenditure per prof)/(revenue per student).
If the average prof teaches 4 courses and the average student takes 10, then just multiply by 10/4 to find the average class size. On average, there must be enough students sitting in the class to pay for the prof teaching them plus the classroom and all the other overheads.
A university can do almost nothing about revenue per student. It's set by the provincial government. Maybe we can lobby. Maybe if we bring in a bit more money in donations that would increase it a little. But basically it's exogenous. (OK, places like Harvard with zillion dollar endowments aside).
A university can do a little bit about expenditure per prof. Lower profs' salaries. Try to cut the electricity bill a little. But basically it's exogenous too. It's set by the market, with a bit of union power thrown in.
So, basically the student/prof ratio is exogenous to the university. The university can choose the number of students, and choose the number of profs, but it can't choose both at the same time. It can't choose the student/prof ratio.
When you hear a prof whining about the student/prof ratio, and demanding the administration hire more profs or let in fewer students, you can be fairly confident he is not demanding a cut in profs' salaries, or demanding profs teach more courses. So you can be equally confident he doesn't have a clue about how the student/prof ratio is actually determined.
OK. Now you understand the most important lesson about universities. Which is something very few profs or students understand.
Next, you have to understand that while the student/prof ratio is exogenous to the university as a whole, it is not exogenous to the individual prof, or to the individual prof's department. Here's where we enter the Hobbesian State of Nature: the war of all against all to get someone else to teach all the students that pay their salaries.
The individual prof wants a lower student/prof ratio. He can try to avoid teaching the big intro courses, and try to persuade the chair of the department to let him teach his own prissy little boutique course, preferably for graduate students, in his own little research area. Or just be strategically incompetent at teaching the large classes. So you see departments having a large number of small classes, taught by the most favoured profs, and a few very large classes taught by the less favoured or powerful. This distorts the distribution of class sizes. It's far too skewed. Plus, there's always the old standby. Too many students? Start doing some really hard stuff, and try to scare off some students to try to get the numbers down to what you think is best. Or just teach badly.
The individual department chair also wants a lower student/prof ratio. He can avoid putting on courses that attract lots of students. He can put limits on the number of seats in a class.
Each individual prof, and each individual department chair, can reduce his own student/prof ratio; but the overall student/prof ratio for the university is fixed exogenously. Their individual attempts must fail collectively. But their individual attempts nevertheless have some very undesirable consequences.
The supply of seats is determined by the individual department. But the demand for seats is determined centrally, by Admissions. Admissions is ordered to bring in as many students as are needed to pay the profs' salaries. But each individual prof and department wants to reduce the number of bums on seats in his course and his department.
The result is a classic case of chronic excess demand for seats. Just like in the old Soviet Union. Capitalist economies have chronic excess supply. Communist economies have chronic excess demand. My job as associate dean (because I made it my job) was to persuade, cajole, bribe, threaten, or bully departments into putting on enough seats for the bums that needed or wanted to sit in them. I was what the Russians used to call "the pusher".
The Chair of Department X only made enough seats available in Intro X for two-thirds the students who wanted to take Intro X. Students already majoring in X got a seat; but a lot of students wanting to take X as an option couldn't find a seat. Some had to take Intro Y instead, even though they preferred X, or would have been better at X.
The Chair of X actually had a fully worked out model which determined his choice of how many seats to create. He knew that a certain percentage of students who took Intro X would discover they liked X and were good at X, and would want to switch their major to X. But, with a given number of profs in X, any increase in students majoring in X would simply result in bigger class sizes in upper year courses. The Chair of X thought those classes were big enough already. So he restricted seats in Intro X to give him the upper year class sizes he wanted. And my job was to try to convince him to act in a way contrary to his own department's interests. I would argue that, in the long run, the number of professors he had was proportional to the number of students he taught. But I had no hard evidence to support this assumption. And he was worried about student/prof ratios right now.
The university as a whole faces a hard budget constraint. Individual departments face a soft budget constraint. Individual profs face none.
It's really strange trying to allocate resources efficiently where there's generalised excess demand.
You get gridlock, where one student has a seat in course X, wants a seat in course Y, but can't find one. And there's another student in course Y, who wants a seat in course X, and can't find one.
You get hoarding. Students grab a seat in any course that's open, even if they think they will probably not want it, just in case they do want it and can't find a seat later.
What's worse is that you get Clowerian spillovers. Walras' Law is totally false. The sum of the excess demand for seats can add up to anything whatsoever, because the student who can't get a seat in course A, will try to get a seat in course B instead, and if he can't will try for course C,...then maybe try course A again, and so on. So the central planner (that was me) trying to find out how many extra seats in which courses are really needed, hasn't got any idea. A notional excess demand for one seat in one course can create an effective excess demand for hundreds of seats all over the university. I am one of the very few economists who has actually seen the Barro Grossman supply-side multiplier at work, in real time. Create a couple more seats in one course, and seats suddenly start appearing in other courses all over the university.
The very worst case I saw was when Barney was cancelled. Barney, a very large first year course, about dinosaurs or something, was one of the most profitable courses in the university. And it was cancelled, by the department.....to save money. There was any number of prissy little boutique courses that could have been cancelled. But they cancelled Barney. We had to go all the way up to the VP Finance to get Barney a reprieve.
The problem is obvious. Barney made tons of money for the university. But the individual department saved money by cancelling Barney, since nearly all the students taking Barney came from other departments and faculties. The incentives just didn't line up.
And the solution is obvious too: make the money follow the students.
I first saw the solution in action in our Summer Incentive Plan.
Classrooms were close to capacity in Fall and Winter terms, but well below capacity in the Summer. If we taught more students in Fall and Winter, we would need to build more classrooms, which cost money. If we taught more students in Summer, we didn't. So marginal costs were lower in Summer. So the university set up a Summer Incentive Plan to give each of the four Faculties a share of the fees of any additional students they taught in the Summer.
The Dean of our Faculty was an economist. He decided that all the profits from the Summer Incentive Plan would flow through to the individual departments that succeeded in increasing Summer enrolment. What's more, he said he would underwrite any additional costs we faced in putting on extra courses to attract more students. Departments could win, but could not lose.
I was Chair of Economics at the time. I made the Dean explain it to me twice. When I was sure the Department could not lose, and could only gain money, I looked through the Calendar, and put on every single course for which I could find a qualified instructor. The Chair of Business did the same. We made a killing that Summer, and an even bigger killing the next Summer. So did Business. The other departments finally followed only after they saw that Economics and Business had spare cash to spend on all sorts of nice things, like seminars. And eventually students had lots of choice of courses to take in the Summer.
The Summer Incentive Plan worked very well in two Faculties, and failed in the other two. The problem was the Base Year Effect. The Faculty only got money for an increase in students compared to the base year. The two Faculties which failed to grow their Summer enrolments were stuck in a Poverty Trap, because they had been unlucky to have had abnormally hgh enrolment in the base year. And they got a bit confused about the distinction between average and marginal costs and revenues as well. They wouldn't hire more instructors to teach more courses unless they had the money to do so. And they never got the money to do so because they never put on more courses to attract more students. They never seemed to figure out that investing in an extra course that brought in more than 15 extra students would earn a profit.
But in my Faculty I now saw a big difference between Fall/Winter and Summer. There was never a shortage of seats in the Summer. Departments were hungry for students. They never complained that they didn't have the resources to teach extra students. They knew the resources would follow the students. The obvious thing to do was to extend the Summer Incentive Plan to Fall and Winter as well.
It eventually happened, but it took a long time to hammer out the details and implement it.
Was there to be a base year, so that only enrolments above the base would be rewarded, or should it be zero based? The former leaves existing disparities untouched, but a zero based plan would leave some parts stuck forever in a Poverty Trap, lacking the resources needed for a Big Push.
Should the money flow to Faculty Deans, or right down to Department Chairs?
How do we handle the fact that some students bring in more money than other students?
How do we allocate overheads?
Should we give less money for students enrolled in first year courses, and more for students enrolled in fourth year courses?
If an economics student takes a course in maths, how much money goes to Economics and how much to Maths?
Precisely what expenses would the Faculties and Departments be responsible for covering with their incentive money?
Our ad hoc Committee of Nerds (economists, mathematicians, statisticians, and accountants who knew how the university was run) grappled with all these questions. There were no easy answers. But we cobbled together some sort of workable compromise, and it was implemented last year.
It's far too early to say that it has worked. I'm no longer working as a central planner, or pusher. But the people doing those jobs tell me it has changed attitudes. Individual departments now want to attract students, and get more bums on seats. That's what I wanted it to do. Collectively they will still fail, of course. Because Admissions won't let in any more students. But the students we do let in will find it easier to get seats in the courses they wanted, and will have a better experience. The allocation will be determined more by demand, and less by supply. We will become an excess-supply economy, not an excess-demand economy.
Let me try to list some of the lessons I have learned:
1. An organisation that is officially a central planning economy really isn't. It's a mixture of central planning and Hobbesian State of Nature. This is really just the old Hayek story. Central planners lack the information, and lack the power, to get people to do the right thing.
2. It's hard, in fact impossible, to get prices right. You will always end up over-rewarding some activities, and under-rewarding others. The "market" will always fail.
3. That doesn't mean you should abandon prices. The prices you set will be wrong. But pricing everything at zero will be even more wrong.
4. But it does mean you can't rely wholly on prices. You will still need to order people around in some cases, and rely on their willingness to do the right thing in others. And you need to hold some money back so it can be allocated at the discretion of the central planner. (Funny, after writing this I realised it is the absolutely standard prescription for a mixed economy; you need some government spending and regulation, as well as personal ethics.)
5. Incentive plans aren't incentive plans. They are anti-disincentive plans. A department chair who doesn't give a damn still won't give a damn, because he can't put any of the incentive money in his own pocket. But a workable incentive plan will offset his strong disincentive to do what's right for the university.
6. Incentive plans aren't incentive plans. They are resource allocation methods. Central planners try to re-allocate resources to where they are most needed. So a department that gets more students gets more resources to teach them. All an incentive plan does is to formalise this resource reallocation into a clear commitment, that departments can rely on.
7. It's not enough (in some cases) to put the carrot in front of the donkey. You have to point to the carrot, tell the donkey it is a carrot, and that he can eat it. And work out marginal revenue and marginal cost for the donkey too. And repeat this several times. This is most true for departments in subjects that economists' prejudices tell us are less likely think like economists. All the usual suspects took a long time to figure out the Summer Incentive Plan and respond to it.
8. The biggest problem is the base year. Imposing a zero-based budget would mean a massive initial reallocation of resources, which you can't do; and some units would just give up, stuck in a poverty trap, if you don't give them any extra resources until they catch up with the rest. But using the current year as base means making an existing misallocation permanent.
Fascinating commentary Nick. The implications for resource allocation are very interesting. Just for fun I forwarded to friends at McGill's Physics dept -- maybe they will beat Economics and Business to a similar model there.
Posted by: finance | October 29, 2010 at 09:25 AM
Interesting - a complex process. But, after all of the effort - coming to a sol'n that appears to be working, what would happen if the final allocations of teachers/students/classes etc were then subjected to the same spreadsheet T A&M exercise - and the data made public?
Would it help/have helped the process, or would it throw a whole new wrench into the works? I can see merit in analysing it internally within the university - not sure the sausage making process would benefit from public spreadsheets.
Posted by: Just visiting from Macleans | October 29, 2010 at 10:01 AM
Large firms face similar problems. Within the firm resource allocation starts to look and awful lot like central planning.
Posted by: Patrick | October 29, 2010 at 10:27 AM
finance: thanks! My prejudices say the physics profs will catch on more quickly than most.
JVFM: I think the data should be made available to the profs. And if you make anything available to all the profs, you've effectively made it public.
Even if you don't act on it in any way, making the data public has some good consequences.
First, because the average department and average prof always seems to think they teach more students than average. Dunno why. Maybe one of those standard cognitive biases, or maybe just a Lucasian signal-processing problem, in a world where student/prof ratios are rising, and each prof sees his own ratio but has lagged information on the average ratio. But whatever the reason, it was true in my experience.
Second, because purely symbolic rewards matter too. Research gives you status (as well as money, through outside offers); teaching large classes gives you none. (See Frances' last post and comments). Being able to hold your head up and say you are making a profit for the university gives psychic income, that offsets some of the disincentives.
Patrick: absolutely right. Coase got a Nobel for saying what you just said!
Posted by: Nick Rowe | October 29, 2010 at 11:19 AM
This was really interesting. Why not do like a 4 year average base?
Posted by: jsalvati | October 29, 2010 at 11:22 AM
jsalvati: Thanks! A 4 year average base makes sense. (But central planners think 5 years is always the right number). My memory has failed. I think we did something like that. But all it does is take an average of the problems associated with a zero base and a current year base.
Posted by: Nick Rowe | October 29, 2010 at 11:45 AM
I was thinking the same - internal publishing would eventually become public, through even FOI requests (I think it applies to universities). But, there is a difference to proactively publishing the data.
I could see why some universities would want to be proactive - say one dept head was adamant in not giving up one pet area of study. Nothing that a bit of public scrutiny couldn't easily resolve.
Posted by: Just visiting from Macleans | October 29, 2010 at 12:04 PM
Nice post! I'll bet you're happy to be able to turn the tables for once! But: "The biggest problem is the base year." Since the problem seems to be an unacceptable rate of change, what's wrong with
exp(-at)base + (1 - exp(-at))zero
With suitable choices for reversion speed a and initial time t?
Posted by: Phil Koop | October 29, 2010 at 12:39 PM
Phil: thanks. Yes, it was lovely to be able to turn the tables, and say something to other people that commenters say to me, about practical real-world experience.
Your idea of steadily changing weights so there's slow adjustment towards zero based budget was almost exactly the one I proposed on the Nerd's Committee. But it was ruled out as far too complicated for most of the departments and Faculties to understand. Even very simple carrots are too complicated. My hope is that, in the long run, steady overall growth will make the historic base less and less relevant. So we will end up with something similar to declining weights in practice.
Posted by: Nick Rowe | October 29, 2010 at 01:29 PM
Fascinating.
"This is most true for departments in subjects that economists' prejudices tell us are less likely think like economists. All the usual suspects took a long time to figure out the Summer Incentive Plan and respond to it."
I love this idea of (forcible) expertise-sharing across departments. Are there any associate deans from the English department cooking up a scheme to make the economists learn how to write and speak better? (you, I think, don't need this help but plenty of economists do)
Posted by: Leigh Caldwell | October 29, 2010 at 02:09 PM
Nick, great post! Of course you realize that, in the Texas A&M scheme, your administrative work - which is far more valuable than some prissy little boutique course - would count for nothing.
Posted by: Frances Woolley | October 29, 2010 at 02:41 PM
Frances, in the command economy known as the private sector, we have a solution for that too, namely internal transfer pricing. And yes, we all complain loudly about these monopoly providers!
Posted by: Phil Koop | October 29, 2010 at 04:00 PM
Leigh: Oh God! And the sentence of mine you quote contains a grammatical (and I needed spell-check there, for 2 m's and 1 t) error! It should be "...less likely TO think like economists.."
Yes, we need help from outside. But the English department would probably be talking over our heads, plus not taken seriously enough by us. Many more of us took Dierdre McCloskey seriously, about the importance of good writing (I know I did), even though a lot of what she said went over my head too.
Thanks Frances! Yep. All pricing schemes leave some stuff out. Which is why you either make them incredibly complex, or keep them simple but add a discretionary fund for the stuff that gets missed.
Posted by: Nick Rowe | October 29, 2010 at 04:36 PM
Nick, good stuff. Enjoyable to read these sorts of bricks and mortar articles since they bring it all back to reality.
Faculties sell a product, just like everyone else. Linking their respective share of total firm resources to their sales output probably makes a lot of sense. A teacher can sell more education if they make their course more interesting, remove seating limitations, add scheduling flexibility, or if they offer easier marks. Hopefully not too much of the latter.
Posted by: JP Koning | October 29, 2010 at 04:48 PM
Quite interesting and totally believable. Are there any institutions where research is anywhere as important financially as it is academically?
Posted by: Lord | October 29, 2010 at 05:03 PM
Apropos point 7, Nick, what importance do you attach to the absence of an obvious stick, in the original Summer Incentive Plan, and in the full-year plan? Would people have been more easily convinced that the carrot was a carrot if there was a stick, too? Would the plan have been more or less successful with a stick as well as a carrot, do you think?
Posted by: Greg | October 29, 2010 at 05:07 PM
There are two types of profs, tenure/tenure track profs who are extremely costly to fire, and sessional profs who cost a few thousand dollars per course and can be hired and fired at much lower cost.
How many students does it take to pay the salary of a sessional prof+other related expenses? Isn't the number in the 10 to 15 per course range? So when sessionals are teaching 70, 100 students, aren't enormous rents being generated?
Who is capturing them?
What would be the implications of a different division of the surplus created by sessional profs?
Posted by: Frances Woolley | October 29, 2010 at 05:57 PM
What a beautiful post!
A long time ago when I first entered univ. I used to have debates about this with my prof. In the community colleges, class sizes are smaller, and professors teach 5-6 classes. At the university, they teach two classes -- to free up time for research.
Simultaneous to that, they hire lecturers to teach the intro classes. And the cost of the hiring the lecturers exceeds the cost of the research grants obtained. A big state school like Texas A&M is probably getting only 20% of their revenues from research grants.
But 100% of the faculty hiring decisions are based on research, and the relative pay of the lecturers vs. the faculty reflects that priority.
At the time, the explanation was that some revenues are not counted -- "sunk" revenues? For the department and also in some sense for the school, the teaching revenues are exogenous, but the research revenues are not. In that way you can spend a billion to build a public institution, and then someone else comes along and dangles a hundred million in front of that institution, and the whole enterprise re-orients itself at getting the additional revenue. Now you are spending a billion to buy a hundred million, and marginal revenue is less than marginal cost.
Posted by: RSJ | October 29, 2010 at 06:05 PM
very interesting post... i wonder if this has the effect of reducing the number of terms it take for a student to get through the system - since students have more choice and are less likely to be excluded froma course they want or need in order to do some other course, but is not part of their major.
i would think that a 4 year average, starting with the current year, would make sense, and then factor the average from 100% down to 90%, particularly if admissions grow.
the real problem i have with canadian universities and government planning is this: we have a shortage of doctors - 2.2 per 100, while the oecd average is 3.0. rather than increasing the number of slots in medical schools, the focus has been on immigrants and "credentials recognition" instead of increasing the number of graduates by 50%.
the provincial governments prefer the status quo, as a means of rationing healthcare indirectly, by limiting doctors. the real challenge should be to get some faculties to reduce the number of students taking general arts or things that are not in demand (generaly, those who atrract the students with the lowest high school marks is a good indicactor of market demand) while increasing the number of people graduating from those faculties that have students with the highest marks going in.... medicine, engineering, etc.
Posted by: btg | October 29, 2010 at 08:08 PM
I have to say the idea of having the money follow the students sounds good.
Year 1: Economics department offers dumbed down but popular courses (Freakonomics 101 - why driving drunk is a better idea than walking home drunk?) and draws in lots of students.
Year 2: Tired of being short of cash, Languages decided to fight back and embarks on a 'the customer is always right' strategy in which all students are awarded 100% for every course. Enrollment booms.
Year 3: Economics department also guarantees all students 100% in every course and in addition starts offering students cash bonuses (down to the point that marginal revenue = marginal cost) to take their courses.
Year 4: All department guarantee 100% grade with no attendance required and cash back to the student.
Year 4: Economist overseeing program sees that all deadweight losses have been removed from the now perfectly competitive market and declares the program a success.
Posted by: Declan | October 29, 2010 at 08:51 PM
JP: trying to get different departments to have consistent grading standards was another of my attempts at central planning. Less successful. An econometrician colleague did an enormous econometric study for me. "Operation Birdhunt". Two-way fixed effects model(?) of grading across the whole university. A dummy variable for each course (or maybe each subject?), and a dummy variable for each student. You have to have a dummy for each student because high grades might mean it's a bird course, or might mean it attracts very good students. So you have to estimate the easiness of each course and the ability of each student simultaneously. The supercomputer actually solved it. But few of the results were statistically significant. Not enough mixing in the data set. Science students not taking enough arts courses, and vice versa, so the data could not estimate it accurately enough.
But I didn't see any evidence to correlate grades with desire to attract/repel students. It was more just differences in local customs across disciplines. Some are softies, some are toughies.
Lord: a university's success in getting research dollars is great for its prestige. But AFAIK, it's usually teaching that brings in the revenue. And undergraduate teaching at that. Grad students bring in about twice the revenue per student (less for MAs, more for PhD's) but cost more than twice as much to teach, given much smaller classes, not to mention the scholarships you need to attract them.
Greg: in most cases, the only stick we could implement was not giving them the carrot. Departments believe they have property rights in resources they have already been allocated under central planning. Plus, it's very hard to fire profs. You can fail to replace them when they retire or quit, but even that is hard, politically. One of the big benefits of an "incentive plan" is that by calling it an "incentive" you get the legitimacy to take it away if their student numbers drop. But an incentive plan that actually had negative payments...no way. That's why we couldn't do zero-based budgeting.
Frances: Remember, we call them "contract instructors" now! The word "sessional" was banned a few years back (not sure why).
Under the old Summer Incentive Plan, with the old contract instructor rates, when I was running the shop, the break-even class size was about 15 students (I used to know it to the first decimal place). The new plan is more complicated. It depends on which students are in the class.
Carleton is very lucky in having a big pool of some very good contract instructors in Ottawa, willing to work for lower rates per class than regular faculty salaries. (And many of them are people we really want to have teaching for us, and who can't be doing it for the cash.
Who gets the rents? Good question. I think the students get most of them. Faculty probably get some too, in the form of teaching smaller classes than we otherwise would have to teach. My guess is that faculty salaries would not be much lower if we had to pay more for contract instructors.
RSJ: Modelling universities as a Yugoslavian worker-controlled cooperative is not far off the mark. Where it's the regular faculty who are the subset of workers who own the firm. And faculty want prestige, and status, and want to be part of a high-status university. And status means research. (Economists don't pay enough attention to status as a motivator, I think).
Posted by: Nick Rowe | October 29, 2010 at 09:07 PM
btg: I once persuaded the university statistician/econometrician to do a study for me: to see if the last 2 digits of the student's ID number predicted retention. My theory sounds totally daft, but it isn't, because we let students register in courses in order of the last 2 digits. Every year we change the order back to front, but one year we forgot. So I had a natural experiment. But he couldn't find any significant difference. (The computer techie who ran the program allocating student ID numbers thought I was absolutely nuts when I phoned him to ask if the last 2 digits had any meaning.)
The government faces one big problem with the current "voucher" system. It is very hard to have a voucher system that promotes competition for students between universities, and at the same time puts an upper limit on the total number of vouchers in a particular subject area, like the BA.
There used to be an upper limit for each university. But that caused a lot of lobbying by individual universities to get their limit raised, and prevented competition between universities. We do have limits at the graduate level, especially binding at the PhD level.
Declan: but remember, all a good "incentive" plan does is remove the disincentive. Departments don't necessarily care about absolute size, only about the student/prof ratio, and a good pricing system should make that independent of size. And departments also want to teach *good* students. So you might think there would be a tendency to water down the courses, and grades, but it doesn't seem to happen.
Posted by: Nick Rowe | October 29, 2010 at 09:28 PM
NR, allow me to be devil's advocate and take your Barney example. You state:
The very worst case I saw was when Barney was cancelled. Barney, a very large first year course, about dinosaurs or something, was one of the most profitable courses in the university. And it was cancelled, by the department.....to save money.
Now, notwithstanding that paleontologists may not be in great demand, how can this be viewed as "one of the most profitable courses in the university"?
Aren't you really saying, that given that a BA requires x many of credits, (and overall admission levels were predetermined before actual course selection by individuals) this was the cheapest way of delivering one credit? I can expound further.
Posted by: Just visiting from Macleans | October 29, 2010 at 09:45 PM
"Departments don't necessarily care about absolute size"
They don't? I'm also skeptical if we really know whether or not courses are being dumbed down and what the contribution of various campus programs might be to the educational standards.
But I agree that if you can remove perverse incentives then that is all to the good. I was just highlighting some of the potential downsides. Your story reminded me of this New Yorker article about the U.S. country with the highest health care spending in the country and the risks of introducing monetary incentives into systems with non-monetary goals and of introducing competitive pressures into situations where universal standards need to be maintained.
Posted by: Declan | October 29, 2010 at 09:57 PM
JVFM: One prof delivered one credit to a very large number of students, with Barney. Barney was/is a very popular course. A large number of students were prepared to spend their fees and their vouchers on Barney.
Posted by: Nick Rowe | October 29, 2010 at 09:58 PM
Declan: remember that Admissions still controls entrance requirements centrally.
Central planners used to have to resist pressures from individual departments to raise admissions requirements in their particular programs, to force the weaker students into other departments to pay the bills, while they taught only the good students. There exists some optimum level of requirements. It's not infinitely high, nor infinitely low. The aim should be to align the requirements that are good for the individual department with the requirements that are good for the university. They used to be out of alignment. If the individual department faces the same budget constraint as the university as a whole, it ought to bring them closer into alignment.
But ultimately, we will see.
Posted by: Nick Rowe | October 29, 2010 at 10:09 PM
Yeah, I get that. But one way to look at organizations is to classify them as cost centres, revenue centres, profit centres.
Overall, the university is a revenue centre (# of students x annual tuition)
Individual depts/courses are cost centres.
Anyway, more a semantics argument - not really important.
Posted by: Just visiting from Macleans | October 29, 2010 at 10:19 PM
Admissions controls the standards to get in, I was talking about the standards to get out :)
The risk is that departments don't face the same constraint as the university because getting a student to move from English to Economics could help Economics and hurt English, but the university as a whole could either be better or worse off depending on what caused the student to change departments.
Posted by: Declan | October 29, 2010 at 10:29 PM
Declan; agreed. It *could* happen, if departments get too hungry for students, that they will compete against each other to lower continuation standards below what the university wants. I haven't heard of it happening yet. If it does become a problem (I don't think it will), one response would be to lower prices, to make each department a little less hungry for students. Because it would mean we have ended up overshooting our target.
Posted by: Nick Rowe | October 29, 2010 at 10:59 PM
Nick:
Your experience in Canada seems much different from mine. For instance, you write: "Nearly all revenues come from teaching students, and are proportional to the number of students taught."
Disagree. The top source of revenue comes from the overhead assessed on research grants. Tuition dollars are valuable though because only certain kinds of costs can be paid with grant overhead: the buildings, the professor's base salary, secretarial staff.
So for instance, Caltech had a 3:1 student faculty ratio. Benefit? The faculty are required to teach only one class per term and only for two of the quarters a year. But as your math suggests 20K tuition per student is about half the average faculty salary, and we haven't even begun to pay for anything else! So in short, there is a high ratio, but professors don't teach much--normal course load is 5-6 classes/term.
So is there a lot of competition to get into classes? No. This problem is solved by not having any course size limits. What does happen is that the number of teaching assistants goes up. Even after the class has started, if enrollment continues to grow.
AFAIK, the faculty and the departments love this system, for reasons you suggest. First, money from the tuition is distributed to departments to defray the cost of teaching, and as you suggest this gets turned into food and drink :). Second, the funds for teaching assistants increase. So this subsidizes the graduate students.
Now all the professors supply RAs out of their research grants, which includes cover for tuition, but if the student TAs two classes a term, the department picks up the entire tuition portion. So less drain on the professors. The graduate students like it because its usually possible to pocket double pay, once from being an RA and once from being a TA.
The icing on the cake is that many grants allow the professors pay themselves a supplemental salary out of the grant money directly. So more students in the class -> more TAs -> more 'extra' money in the grant. More TAs -> easier to support more graduate students -> more grants. Etc.
This pattern isn't unique to Caltech. I've been an adjunct at USC, and many of the professors don't teach at all. They get waivers because their research grants are so large. Then the department turns around and hires someone like me.
Its a huge money making machine. I've taught classes of most 50-80 graduate! students. Tuition is high and unsubsidized for master's students. So that works out to be about 500K in revenue for a single class.
Posted by: Jon | October 30, 2010 at 12:52 AM
Jon: yes, that does sound like a very different university. My little model wouldn't work very well there.
Do I understand correctly that a prof or department that gets an increase in the number of students *automatically* gets a proportionate increase in the number of TA's and contract instructors to help teach them? So there's an incentive plan (aka resource allocation mechanism) that makes the money (aka resources) follow the students? That's what's important.
And if a university can set its own fees, then it presumably faces a downward-sloping demand curve, and so revenue per student is not fixed exogenously for that reason as well.
Posted by: Nick Rowe | October 30, 2010 at 07:38 AM
This was most interesting, but like Jon, my university also receives revenue from research grants plus paid grad assistants written into grants, and, not to be underestimated, the reputational benefits that accrue to the university from having faculty doing funded NIH/NSF/CDC (to name a few) research. Those of us who bring in grants necessarily have lighter teaching loads. Interestingly, at my previous university, the claim was that they lost money on research (they recovered about 65 cents in indirect costs on each direct cost dollar from federal grants) and did it only for the reputational benefits. As an economist, I found this implausible, at least in the long run. The reputational benefits must at some point translate into revenue (more students, more patients at the med school, something) to justify the costs and if they generate revenue in the long run it must (eventually) equal the "costs" associated with the research. Otherwise, why do it? Ergo they aren't (or won't be or believe they won't be) losing money on research. When I suggested this might be the case, it was always asserted that "no, we lose money on research." Weird business model.
Posted by: Maxine Udall (girl economist) | October 30, 2010 at 08:42 AM
Welcome Maxine!
If research grant pays for the teaching release, the overheads the university skims off the research grant would seem to be mostly profit for the university.
To my mind, university "reputation" shifts the demand curve of new students to the right. And the university then has a choice of whether to accept more students or raise admissions average or reduce scholarships or (not in the Canadian case) raise fees. My guess is that the people saying they lost money on research might have been saying they "spent" the reputational benefits by raising admissions standards. And they ignored the fact that this saved them money from not having to lower fees or increase scholarships to get an equivalent increase in admissions standards.
Posted by: Nick Rowe | October 30, 2010 at 08:55 AM
"or (not in the Canadian case) raise fees."
Sometimes there are brief windows of fee deregulation in Canada. U Waterloo took advantage of one of them to set their tuition for undergrad accounting and accounting/math combined programs at $12,000 per year, more than double the fees charged by comparable programs elsewhere (Waterloo, of course, would argue that there are no comparable programs elsewhere)
Interestingly, though the combined math/accounting programs charges the same $12,000 fees as the straight accounting program, math gets none of the extra revenue - at least this is the gossip I've heard on the street.
Posted by: Frances Woolley | October 30, 2010 at 10:53 AM
Maxine writes: "When I suggested this might be the case, it was always asserted that "no, we lose money on research." Weird business model."
In Biz school, there are two methods of accounting - the bean counter Generally Accepted Accounting Principles(GAAP) which has all sorts of dated rules on how fixed costs are allocated (often against labour) and Management Accounting - which would be more in line with the way economists look at things.
In a simple example - a research grant of $100k, prof salary and expenses $90k. The university is ahead $10k.
Add in allocated fixed costs (which do not change) due to GAAP accounting convention against labour (say 20% of prof salary and expenses) and the "cost" goes up $18k = $108k. The university is "losing" money on research.
Always be cautious about how overheads/fixed costs are allocated.
Posted by: Just visiting from Macleans | October 30, 2010 at 11:38 AM
Btw, the situation that Jon describes is more inline with a "profit centre" - you control inputs (grant money and possibly fees charged) and indirectly costs - number of TAs/splits to Profs- so profit maximization has a different dynamic.
In a "cost centre" - like a factory - the dynamic is different - so if you're GM and you're running full out with two production lines/overtime on weekends at time and 1/2 Sats and double time Sundays - it eventually becomes cheaper to run three shifts M-F. Avg cost goes down.
Posted by: Just visiting from Macleans | October 30, 2010 at 11:57 AM
Yes. The number of TAs is automatic. At Caltech, they mostly do not allow contract instructors. So no to the second point. USC uses lots of outside instructors, and yes in that instance.
I found a to illustrate how tilted operations are:
(Annual) Revenue:
Tuition: 19M
Research Grants-direct: 174M
Research Grant Overhead: 100M
Auxiliary (Housing,etc): 34M
JPL Contract: 1.6B
Investment Return: 214M
Gifts: 27M
...
Expenses:
Instruction: 221M
...
Posted by: Jon | October 30, 2010 at 12:38 PM
Frances: this is how I think fees work in Ontario:
Foreign students: we can charge anything we like, and get no money from the government.
Canadian students: we can charge the official fee, and get a voucher. Or we can charge above the official fee, and get no voucher.
JVFM: Yep. I had quite a job arguing that we should not allocate fixed costs of classrooms to the Summer, when we were well below capacity.
Jon: Wow! That's all I can say.
Posted by: Nick Rowe | October 30, 2010 at 02:06 PM
The problem with Nick's central plannig tendencies begin when he wants the central bank to create money to prop up the value of equities.
That's central planning, unconscious of malinvestment--the sort of thinking that got us into a stupid real estate bubble.
Central planning has to be done with an explicit distributional benefit, if it is to be done at all. Reflation by itself is not a worthy objective for the central planner, and if pursued, will result in a political, as well as economic, correction.
Posted by: Roland | October 30, 2010 at 02:49 PM
Say, one other q about the summer program (not surprising economics and biz depts were the first two to subscribe). Did this actually result in an increase in total university enrollment, or were you just getting a larger slice of the fixed overall pie (student-credits)? If it's the latter, shouldn't one expect both winners and losers - take from Peter Dept to pay Paul Dept?
Posted by: Just visiting from Macleans | October 30, 2010 at 03:28 PM
JVFM: there was an overall increase. It's difficult to say whether different departments' courses are substitutes or complements. Probably substitutes on balance. But for foreign students especially, if they know there's a good variety of courses being offered, it can make it worthwhile to stay full-time in the Summer.
Posted by: Nick Rowe | October 30, 2010 at 04:36 PM
With $1.6 billion revenue for the Jet Propulsion Lab (JPL) at CalTech (whose costs aren't really stated) whose research output I presume is principally used for the military (F35 anyone?), one has to wonder whether to classify it in the same category as Canadian Universities.
Posted by: Just visiting from Macleans | October 30, 2010 at 05:19 PM
Just visiting:
I kept the JPL funds on an independent line because its is somewhat different than the other numbers, but no, JPL is not a military oriented organization. They do almost all of engineering and the science programs behind the robotic space missions.
Those funds don't really bleed through to paying professors, students, etc. They subsidize certain services like the library, but I think some government overseers might get upset if too much of the money spilled over into unrelated activity. What it does do is enhance the reputation and provide lots of unique opportunities for collaboration.
Posted by: Jon | October 30, 2010 at 05:58 PM
Jon, notwithstanding that it is for robotic space missions vs military, the relative magnitude of the revenue is what cost my eye - $1600 M JPI/ $2168 M total revenue or 73% of total.
Are you a college with a research facility attached, or a research facility with a college attached? There are synergies, no doubt, either way. Tuition seems to be a rounding error.
Posted by: Just visiting from Macleans | October 30, 2010 at 07:02 PM
Off topic:
I half expect you may be posting on this, because he raises a distributional theme that you've done previously:
http://krugman.blogs.nytimes.com/2010/10/30/accounting-identities/
Although he's not entirely accurate when he says:
"Now, one person’s liability is another person’s asset, so rising debt made the world as a whole neither richer nor poorer."
That's a sub-story myth about the flow of funds that I'll refrain from commenting on.
Posted by: JKH | October 31, 2010 at 08:41 AM
I work for a large diversified privately-owned engineering firm, and I see many parallels with universities. Large engineering firms are also divided into many specialised departments, we also work on large complex projects that may take several years to complete, and we need to attract both new recruits, new clients, and new contracts, based on our reputation for competance (using references based on past successes).
I see 2 difficulties with analysing individual professors (or engineers) on a year-by-year basis:
1) Engineering firms (and universities) suffer from the "hardware store" problem, i.e., thay make a profit on maybe 20% of their inventory, but to draw in customers they have to maintain a very diverse inventory. Engineering firms (and Universities) benefit from the diversity of the skills of their personnel.
2) Junior engineers do not become profitable until after several years, and are most profitable in their middle years. By their waning years their productivity drops of in a strictly technical sense, but they benefit the firm in other ways by bringing in customers because of their long experience which lends credibility to service offers, and by helping out junior engineers with some of the more subtle "soft" problems requiring judgement and diplomatic skills.
Posted by: Alex Plante | October 31, 2010 at 01:49 PM
JVFM: "Tuition seems to be a rounding error."
That was worth a good laugh from me!
JKH: Funny thing is, about a week or so back, PK had a post which said the private sector couldn't pay down debt unless the government added debt. And I was thinking "that's true for net debt (ignoring foreigners) but not true for gross debt, because the creditors could spend more and the debtors could spend less". Then he did the most recent post, making the same point.
"Although he's not entirely accurate when he says:
"Now, one person’s liability is another person’s asset, so rising debt made the world as a whole neither richer nor poorer."
That's a sub-story myth about the flow of funds that I'll refrain from commenting on."
I couldn't see what you were getting at here, at first. Then I realised of course you were right, because if A has real investment projects with a higher rate of return than B's investment project, B lending to A could make both B and A richer. That's one of the main things that debt is supposed to be about. That's the main reason we have a financial system.
Is that what you were thinking too?
Alex: Similar problems to your 1 and 2 crop up in universities as well, though perhaps not so severe. But that's one of the reasons we didn't try to push the incentive plan down too far, to the individual professor level.
A similar problem is this: profs can be hired, but not fired. The only quick flexibility comes from contract instructors. If student demand fluctuates too quickly, from one department or subject to another, it becomes impossible to make the resources follow the students.
Posted by: Nick Rowe | October 31, 2010 at 04:13 PM
Nick,
Simple illustration of what I’m getting at:
Suppose a closed economy generates I = S of investment and saving. Leaving aside marked to market, that becomes a measure of accumulated wealth or “riches” over time, technically speaking.
Suppose for a given period there is no internally generated equity in the corporate sector (i.e. no undistributed profit – could still be dividends or not). That means all of S comes from somewhere else.
Assume investment I takes place entirely in the corporate sector.
The corporate sector then has to finance externally all of its investment I.
Assume it finances entirely with debt.
Then the entire wealth accumulated in that period is represented in finance form as debt (either directly held or indirectly held by the household sector.)
So it is not necessarily true that “rising debt made the world as a whole neither richer nor poorer”. And in fact, over the long haul, it hasn’t been true at all. The simple example is extreme, but the effect certainly occurs at the margins – and in a very significant way in the US flow of funds accounts. I could take you through that at some point, if you’re interested.
Posted by: JKH | October 31, 2010 at 06:24 PM
JKH: right. Wealth is real capital, and in your (simplified) model, all real capital is debt-financed. So the two rise and fall together. And that simple model contains a large part of the truth.
Posted by: Nick Rowe | November 01, 2010 at 09:18 AM
Optimising students enrolled per prof is all very well, but wouldn't you then take a hit in metrics feeding into stuff like the various World University Rankings, with consequent squealing from university bureaucrats that staff and overseas student recruitment will take a hit?
Posted by: Mark_dowling | November 04, 2010 at 09:53 AM
Mark: In my model, we don't *optimise* the student/prof ratio. We can't do anything about the student/prof ratio. It's exogenous (at the level of the whole university).
Posted by: Nick Rowe | November 10, 2010 at 08:21 AM