August 21, 2020

Research productivity is growing;
Teaching productivity is not

Three weeks ago in this publication, Stephen Parker described the challenges facing professions and occupations that suffer from what is sometimes referred to as Baumol cost disease. This is the idea that some kinds of personal services find it difficult to generate productivity improvements beyond a certain point but they must pay salaries that compete with other occupations that can.

A recent OECD report has found that in 13 selected member countries total expenditure on higher education divided by the number of students doubled in real terms between 1995 and 2015, and cites the phenomenon as an example of Baumol.

Australia was not one of those countries, but understandably readers may think something similar must have applied here if it is a valid general theory. The reality is that the timing, extent and effect of cost disease is more complex and more nuanced than first meets the eye.

It is true that average continuing academic salaries have increased in real terms since 2001 in universities. We estimate that they have roughly doubled in nominal terms, which is a real increase of about 50 per cent, depending on the inflation index used. However, other factors were also in play, including greater workforce casualisation and increased outsourcing.

In a 2018 KPMG report, Reimagining Tertiary Education, of which Stephen Parker was the principal author, costing and productivity data from 2011 to 2016 for 37 Australian universities was used to show that the cost of teaching per equivalent full-time student did not rise to a large extent. Neither did it fall by a large margin.

Over that particular period, there was a small positive education productivity growth (that is lowering of per student cost) but for the median university it was only around one per cent per year.

The calculations were made using the Research and Education Efficiency Frontier Index (REEF Index) and are shown below in a graph from the KPMG Report. The change in the frontier (made up of the universities with the most efficient education and research outcomes) between 2011 and 2016 is highlighted.


From 2011 to 2016, education productivity growth was small compared to research productivity growth. Source: KPMG

Similar calculations by Keith Houghton, using UK data over a different but contemporary time period, produce a similar result. A larger dataset over an earlier time period shows more significant real cost increases within the US university and college sector.

This suggests that Baumol cost disease is not always manifest; rather it may be jurisdiction or time period specific, depending on the prevailing appetite to fund higher education and the extent to which institutions or systems can pre-empt its effect through efficiencies or changes in what is offered.

Turning to the OECD Report referred to in the article three weeks ago, it is clear that for the sample of countries included there was a significant real increase in expenditure divided by the number of students, but not all of that spending was actually on teaching or related purposes. Expenditure also supported research. The proportion of total expenditure allocated to research is a key variable in understanding the significance of the numbers.

It is certainly the case that research activity in Australian universities has dramatically changed. Turning back to the KPMG report, the empirical data described shows that the median level of research productivity growth in Australia is around 10 times that of education productivity growth. For some universities, the differential is significantly greater.

Consider, as an example, the productivity growth at the University of Canberra where Stephen Parker was vice-chancellor from 2007 to 2016.

Movement towards the right of the graph represents improvements in education productivity. Movement upwards represents research productivity improvement. Other than for a small number of years, the University of Canberra has a general pattern of improved productivity in both education and research from 2007 to 2016, but the vast majority of this productivity improvement was in research.

While the pattern at the University of Canberra is particularly positive, many other institutions resembled this type of differential productivity growth.

When one captures both research and education productivity, the Baumol cost disease story is not necessarily one of cost ‘blowouts’ or contagion but may be the result of deliberate choices about where to spend money and where to save it.

This sounds like good news, and in a way, it has been, although it did not always feel like it. But Australian universities may now have reached a productivity ceiling, and if the Baumol thesis is broadly correct over time, it means that a crunch may be looming for Australian universities which is further away in some other countries, irrespective of the impact of COVID-19.

There is some evidence of this. Using the REEF Index diagnostic productivity tool, the extent of research productivity growth amongst half of the Group of Eight universities has tapered off subsequent to 2015.

If year after year, most Australian universities have generated productivity improvements to offset the underlying effect of real “Baumol” salary increases, can they continue to operate at an acceptable level of quality and student learning experience when there is no apparent appetite for real increases in funding by government or students?

Keith Houghton is chief academic strategist of the Higher Education and Research Group. Stephen Parker is global lead, education and skills for KPMG.