Funding universities – newish ideas from the IEA

There is an admirably bonkers new paper from Peter Ainsworth of the IEA with the title “Universities challenged: funding higher education through a free-market ‘graduate tax’”. I’m obsessed, amongst other things, with ways of funding higher education, so I read it cover to cover.

Some background first, as many people will not be familiar with the IEA. They describe themselves as a think-tank whose members, “all those associated with the Institute support free markets – though with different “schools” of free market economics being represented”. Their official history begins with a quote from John Blundell, their former director-general thus “Hayek advises Fisher; Fisher recruits Harris; Harris meets Seldon. In nine words, that is the start of the IEA.” This gives a pretty good idea of their perspective on the world. It isn’t mine.

The report itself, a concise 52 pages, has a good deal of useful material. There is a very good summary of how the UK got to their current situation, where a student loan scheme, intended to reduce Government payments to Higher Education, is now likely to cost more than the direct grant system it replaced. There is a very lucid analysis of the different economic perspectives of students, higher education providers, and the state. There is also a good description of the various perverse incentives in the current system in the UK. There is nothing on the perverse incentives on other countries, for example the US. There is less on the perspective of the families of students, but this is, I think, a relatively minor omission. There is a good review of the graduate premium to earnings, with useful data showing how it has varied over time, and between disciplines. There is particular emphasis on the variation between individuals in this premium, and the consequent uncertainty about investing in Higher Education.

Ainsworth’s key idea is this ” Universities should individually or collectively offer contracts to their students, who would agree to pay to the university they attended a given percentage of their [future] earnings. That percentage could vary by course and institution, though some agreement between universities could be helpful to achieve standardisation. Essentially, the university would be taking an equity interest in the graduate premium earned by the student, although any student who chose to do so could, alternatively, pay the full fees up-front prior to beginning their studies.”

He suggests that, based on a few models around the world, this might be between 4% and 10% of their income over a certain level, for a number of years, and that these payments should receive tax relief. This would be a private contract, and, as such, universities could securitize the earnings.

What’s right with this idea?

  • It could do a good job of aligning the economic interests of students and universities. Presumably institutions whose graduates were of very poor quality, as perceived by employers, would do badly. There would certainly be a strong incentive for universities to give students access to skills and competences, rather than information. Some of these would be specifically vocational, and others more generic.
  • Higher education would be free at the point of use, at least in the sense that there would be no upfront fees. These are, to say the least, a disincentive to poorer students. Ainsworth suggests that the state might choose to provide maintenance grants to support some students at university.
  • It would do a very good job of sharing the risk in choosing to do higher education between the students and the university, with the university taking all the downside risk – i.e. the student never gets a high enough salary to start paying.

What’s wrong with it? Ainsworth does try to address several challenges, for example, there would be an obvious incentive for universities to concentrate on subjects with a high graduate premium – putting less vocational course at risk. Such courses could instead be state supported, as a matter of public policy.

I see some further serious objections, which are not fully considered.

  • Incentives to select students – there would be a huge incentive to the university to cherry pick students, and I foresee a large premium on very low risk people, for example white male students from UK ‘public’ schools (i.e. private schools in other countries). Students with disabilities would be a significant risk to the university.
  • Incentives to select the courses provided – many universities, including all those I’ve worked in, provide courses which are actually uneconomic, or at best very very marginal. We also do a lot of experimentation, running course for a few years to see if they meet a need. These courses, and the associated innovation, provide a wider range of choices for our students, and support a wider range of voices and perspectives within higher education. I doubt that targeted government subsidies would, or could continue to support these.
  • Cost – the cost of collecting these funds is not discussed, apart from a suggestion that universities might get together to do it, or set up a common organization to collect the money. This is not unreasonable, but I suspect the costs would eventually eat up a sizable fraction of the money collected, perhaps 20% or more. The cost of collection is a critical part of any higher education funding system.
  • Risk – bringing in such a scheme is fraught with risk. There would be very large uncertainties, which would put at very serious risk any but the most well endowed (i.e. highly capitalized) of higher education providers. The education system requires fairly stable funding, if it is desired to provide a stable education. A reasonable estimate is that is takes 5 to 6 years from the start of development, to the first graduation of student from a new 4 year degree program. It might be another decade before the graduate premium from such a program could even be estimated. It would be very hard to make a business case for such an investment. The transition costs of any major shift in education funding are large, and are not considered by Ainsworth.
  • Feasibility – Ainsworth argues, on the basis of a number of examples, that such a scheme is feasible. The examples he gives are for small scale, highly targeted schemes. I’ve reviewed each of them, as best I can, and I remain very dubious of the ability to scale such ventures. One important example, CareerConcept AG seems to have had no media activity since 2011, which is surprising. Lumni looks much more promising, but has still only covered 5,000 students in four countries since 2002.

In short, I think the IEA report well worth reading, but I do not think that the proposals in it are feasible, desirable, or affordable. The UK desperately needs a better system of funding Higher Education. This isn’t it.

Using OECD data on health care expenditure

Graphs of OECD data

These notes are mainly to remind me how to draw graphs which use OECD data on health outcomes, health care expenditure. like this one :-


OECD data on total health care expenditure by year across a selection of countries
OECD data on total health care expenditure as a percentage of GDP by year across a selection of countries

 The data come from the OECD stats library, which still requires a subscription. You can download data in several formats. These include Excel files, and compressed comma separated variable (CSV) files. The Excel files use ‘..’ as a missing value character, and need a bit of work to remove formatting, empty columns and rows, and the like before use. The CSV files are compressed with gzip, and I’ve found problems uncompressing them on Linux. The code below seems to work though.

gzip -dfc OECDdata.csv.gz > OECDdata.csv

I use RStudio, and usually use knitr. The libraries I load are here :-

Load necessary libraries

OECD data come in wide format, with each year's data in one column, and the variables which explain it are lined up in columns beside the years like this :-
 "Australia","% gross domestic product",7.6015,7.7003,7.8903,7.8959,8.1065,7.975,7.9824,8.0563,8.2678,8.6248,8.4604,8.5505,"NA","NA"
 "Austria","% gross domestic product",9.4401,9.548,9.6235,9.7954,9.9059,9.8669,9.7359,9.7435,9.9469,10.5356,10.4834,10.2371,10.4102,"NA"
 "Belgium","% gross domestic product",8.1206,8.2921,8.4618,9.6474,9.6751,9.6472,9.5809,9.6237,9.9428,10.6547,10.5577,10.6107,10.8944,"NA"

There are 35 rows, and 16 columns in this particular file. This format is not well suited to graphing, so I need to melt it.

#Total current expenditure as a % of GDP
TCE TCEm'Year')
#Changes values of year from X2000, X2001 etc to 2000, 2001 and so on.

What does this do? The final result looks like this :-
 "1","Australia","% gross domestic product",2000,7.6015
 "2","Austria","% gross domestic product",2000,9.4401
 "3","Belgium","% gross domestic product",2000,8.1206
 "4","Canada","% gross domestic product",2000,8.3075

There is one row for each combination of the two id variable ‘Country’ and ‘Unit’ data point. Each row in the original data set leads to 14 rows, one for every year, in the molten data set.

To make this picture I used this code:-
 # Total current expenditure as a percentage of GDP by year
 g geom_line(aes(x=Year,y=value, group=Country,colour=Country),alpha=0.4) +
 geom_line(data = subset(TCEm,TCEm$Country %in% Ireland),
 aes(x=Year,y=value, group=Country,colour=Country)) +
 geom_point(data = subset(TCEm,TCEm$Country %in% Ireland),
 aes(x=Year,y=value, group=Country,colour=Country)) +
 geom_line(data = IrelandAdjTCEm,
 aes(x=Year,y=value.x.revised, group=Country,colour='red')) +
 geom_point(data = IrelandAdjTCEm,aes(x=Year,y=value.x.revised, group=Country,colour='red')) +
 xlab("Year")+ylab("Percentage of GDP") +
 ggtitle("Total health expenditure against Time")+ scale_colour_hue(c=100,l=50) +
 annotate('text',x=2009,y=12,label="Ireland-GNP") +