Friday, March 03, 2006

University costs

Today I attended a seminar given by Baiding Hu, 'Estimation of Cost Efficiency of Australian Universities' (co-authored with Jocelyn Horne). It uses stochastic frontier analysis to rank the cost efficiency of Australian universities. The costs considered were total spending less estimated research spending and the output measure was student numbers. The most efficient university maximises output per dollar of inputs and then other universities are ranked in terms of this. As studies of this type go this is not too bad and it does include a good bibliography and survey of earlier studies (they skip the recent, Carrington, R., T.J. Coelli and D.S.P. Rao 2005, “The Performance of Australian Universities: Conceptual Issues and Preliminary Results”, Economic Papers, 24, 145-163).

Among the conclusions were that regional universities tended to be the most cost-efficient universities and the ANU is the worst. My own La Trobe University is about mid-range.

I don't believe the conclusions of this study because, like all similar studies I have come across, there are basic measurement difficulties. For measuring teaching outputs one really needs to estimate the value-added to students as a consequence of education in terms of individual cultural or vocational skills. Just measuring student numbers is measuring an input not an output. At La Trobe University, for example, we take students who do not have the highest VCE entry scores but who, upon graduation, often go into jobs that are as well-paid as graduates from larger universities with higher entry scores. We add more value so, if our costs were the same, our efficiency would be higher in this case.

Earlier studies of cost efficiency also examine measures of research funding as research output when research funding is really an input to research not an output. For example, much Australian Research Council money is wasted - it is largely allocated on the basis of ex ante promises not ex post outcomes or measures of previous research success.

While I understand that, with budgetary stringency, emphasis is necessarily on costs, the issue of increasing revenues deserves emphasis. Increasing university incomes by exporting our vocational academic products to new markets is one approach but increased public funding is vital if we are to sustain arts and science programs that produce knowledge as a public good that is inevitably undervalued in markets. The latter is a politically difficult undertaking but this is no reason for not giving it the emphasis it deserves.


P.A. Coplay said...

While the issues of costs and revenues are important - particularly for those working with connections to administration in the sector the study I would really like to see is the one that makes a serious attempt of measuring (even some element of) the public good aspect of education.

For example, If we could identify large changes in the supply of the public good, should we be able to see offsetting changes in the quantity consumed of other goods that are close substitutes for the public good from education?

hc said...

p.a.c., As non-marketed goods it is difficult to come up with a valuation. How do you value the capacity of a society to make better social decisions or to be less narrow-minded? How do you value the extra pleasure people gain from acquiring an interest in literature. It is difficult because willingness-to-pay is not revealled. Maybe non-market techniques such as conjoint analysis can place valuations on different mixes of academic products. I am not confident I would accept the results of such exercises because I think that areas such as the arts are systematically undervalued. I think the arguments probably have to be fought via the ballot box.

P.A. Coplay said...

I agree that trying to directly measure the benefit of the consumption of a public good is going to be extremely difficult (and have even less faith in conjoint analysis to capture it - the effects are much too complex) - however, my impression (which may be wrong as it is not my area) is that there is no even approximate idea of the impact of the public good element of education. So even a crude idea would be of interest. Furthermore my interest is in the benefits to those who are not directly related to the university - as this is important for whether public funding through general taxation is appropriate for tertiary education. if the public good extended only across those directly connected to the university, then this would be an argument for charging those connected to the university more than their private benefits - but no one else.

My rough idea is as follows. While we cannot directly measure the public good component, in a standard consumer choice model, except under some circumstances, a change in the quantity of the public good should lead to changes in the consumption of other goods - my guess is the largest changes will be observed in the goods that are close substitutes for the good. So, if for example, we increased the supply of the public good enormously we should see changes in the consumption of those that are not directly affected by providing the public good e.g. people who are not students or employees of the university. If we observe no change in their consumption then this would be evidence that the public good does not extend beyond those directly related to the university (and of course if we do see changes this would be evidence in favour of broad public funding)

hc said...

You can collect aggregate economic data and see how things like growth rates are related to education and technical knowledge. We know the Solow residual is large. You might be able to examine whether expenditure on non-vocational education has public good effects that increase income or growth or welfare loosely defined. But I just think that as a society we derive direct benefits from having knowledge of our history and from creating good writers and film producers who can help us understand the world about us. I think of Les Murray the poet who has never made much money but who enriches us all with his verse. I am going to post on Les soon as I have just read his biography and like his poems.

Tanya Price said...

P.A. Coplay: Can we interpret from your comments that you reject the last fifty plus years of work on what is termed the social or public returns to higher education? I agree with Harry Clarke that much of this research does not include the more significant yet difficult to measure social returns to public investment higher ed, however the economic returns have been very well measured and tested. Yes, we need to keep on developing better measures, collecting more data, providing researchers with access to that data (now wouldn't that be amazing in Australia?!) and improving our methods of analysis, but the picture is reasonably clear right now.

As far as your general comments on the comparability of government investments in higher education vs other areas of investment, I agree, we do need more research in this area. There are a couple of factors that hinder the production of good quality research in this field. One is that in research-oriented universities, education and research are fundamentally entwined. Any research on comparative benefits has to deal with this social reality (with thanks to Wilhelm von Humbolt for his fathering of the modern professorial role). And any researcher who thinks they have obtained clean/separate measures of research and teaching costs is either someone who truly deserves a Nobel or has not understood that the nature of university finances.

Harry Clarke: I'm struggling to gain any productive insight from Horne and Hu's paper other than - perhaps - further testing of the viability of an econometric method. Their measure of cost efficiency is too crude by far. To more accurately capture cost efficiency, they would need to include not only genuine output measures but also measures over decades. Education's effects not only diffuse but long term.

Though it will be long time (if at all) before we being to capture data of this quality, I would rather see panel studies of students that capture effects of family background amongst other factors. Education is one area where it is very clear that not all inputs are equal.

Am I missing something about this paper?

hc said...

Tanya I agree. In defence of Horne/Hu the task is difficult but I don't believe their conclusions -nor of those numerous authors before them who attempted the same sort of exercise. The main benefit of their paper was to set people thinking about a better approach.

And I think that 'social rate of return' studies which analyse the sustained increase in income-earning potential consequent of education is a start. This value-added must be measured over a lifetime. A literature on rates of return to higher education suggest high returns. These returns need to be related to particular institutions to assess cost productivity.

A difficulty is that much of this 'social return' literature is macroeconomic - it would need to be turned into assessment of research productivity at particular institutions.

On assessing research productivity and non-narrowly defined community benefits, Well p.a. coplay has some ideas but I am unsure. I am certain however that measuring research productivity by looking at research funding as an output is foolish.

P.A. Coplay said...

Thank you Harry Clarke and Tanya Price for your comments - which are most interesting. Will adress both below. But, first should make one point clear. I do not reject that education can be both personally rewarding (indeed it is one reason why to work in the sector) and also increase individual productivity and therefore the productivity of the economy so certainly would not reject out of hand the results of work tying economic growth to education and research.

However, (correct me if I am wrong) but wouldn't such a result in the macro data occur even if the productivity enhancing benefits of education were private - just in the way investing in a capital asset like a factory or like R&D (and I should say that I do not have any problems with accepting that information like a useful innovation is a public good) can return.

However, am less convinced, in the absence of a good natural experiment, of the results of empirical growth models of being able to separate the productivity enhancements from private productivity improving investments from the public elements (public in the strict sense of a public good (non-excludable etc)).

The second point of clarification is what I am interested in seeing measured - it is the consumption or production benefits to those not directly involved. If we think of graduates (and employees) as being information disseminators (along Harry Clarke's arguments (and probably hard to measure)) or else as university education having increased the probability of innovation this would be one argument - again if there is a good study that directly addresses the micro aspects of the dissemination and benefits to the public (non-excludable etc) elements this would be of interest, and my suggestions are along those lines, if such a study does not exist.

hc said...

Yes benefits will arise privately but a difficulty is to also re public benefits. The private benefits themselves are non-trivial to assess as Tanya points out.

I guess you could assess direct consumption benefits using contingent valuation or other techniques or try to tweak something from what atudents do pay.

I think there must be a literature on this - or work that you can borrow from related fields to make inferences.

Tanya Price said...

P.A.C. I could go into a lot further detail on this. It is a complex, intriguing, and important topic. However, I'm going to suggest that you take a look at a very good summary of the research on social returns:

You might want to take a general wander around the CHERI (Cornell Higher Education Research Institute) website. Ron Ehrenberg's research is very good. You might find more of interest in the Institute's working papers .. as well as the published versions that you will be able to find in various economics and education journals.

The paper I’ve linked to cites only a small proportion of the papers in the field but it captures the sentiment well. In particular it does a good job at explaining the focus on macro outcomes (insufficient and likely completely unobtainable measures). It also references the problems with disaggregating both costs and returns. He touches very lightly on the recent research on improved health incomes and reduced social burden.

What is missing from the paper is consideration of the cognitive / behavioural outcomes related to each of the sources of student finance. Ever since HECS was mooted in Australia there has been a rally cry that students that have to pay don’t feel committed to Australia (ie. social return). It’s a frustrating idea for a number of reasons and smacks of (rarely fulfilled) noblesse oblige, however I do not discount aspects of the argument. There is ample evidence in other cases that deliberate gifting produces certain results. In the rush to put dollar signs behind private and public benefits we seem to have forgotten to understand the opportunities before us. HECS is in many ways a very well designed program but the idea of it being a social gift has not been emphasized. I think that we lose a valuable opportunity when we fail to make that connection. How that happened is a whole other can of worms!