COVID-19 has been bad for jobs in higher education. Last October, the NTEU estimated that since March 2020 more than 12,000 jobs had been lost. According to Universities Australia in February 2021 at least 17,300 jobs were lost in 2020. But how many jobs were there to begin with?

This is a surprisingly difficult question to answer. The official DESE staff statistics give us a headcount as at 31 March each year of people employed on an on-going or fixed term contract at public universities and Bond, Notre Dame, University of Divinity and Avondale University College. At the end of March 2020 these institutions had just over 130,000 employees.

But as the chart below shows, the full-time equivalent count is always higher than the on-going or fixed term headcount, because it includes casuals. On a FTE basis, about 18 per cent of all staff are casuals, including nearly a quarter of academic staff. But DESE does not collect headcount data on casuals.

Casuals

While DESE does not provide a casual headcount, two datasets that include higher education providers, produced by the Australian Charities and Not-for-profits Commission and the Workplace Gender Equality Agency, include casual figures.

For the ACNC, which has 2018 as its most recent year, the figures are from the last pay period of the reporting year. University data is calendar year. They report 126,034 on-going or fixed term university staff and 97,561 casuals, making casuals 43.7% of staff out of a headcount total of 223,439 (Torrens University omitted; in some cases the ACNC counts staff in university affiliated entities that are excluded from DESE figures).

This and all the casual headcount data I cite over-states the number of distinct persons, as some casuals work for more than one provider. An NTEU survey in 2015 found one-in-five casuals worked in more than one higher education provider, although not necessarily simultaneously.

WGEA has published data up to 2019-20, with the data described as from a snapshot date between 1 April of the previous year and 31 March of the current year. Organisations can choose any date but it should be ‘representative of your workforce for that year’. This creates the possibility that Australian higher education institutions have chosen different dates, possibly in different years.

However, sampling the first ten universities in the WGEA spreadsheet in seven cases their reported 2019-20 numbers varied from their DESE 31 March 2020 figures by less than 1 per cent. The one significant variation was Swinburne, but this appears to be because Swinburne associated entities are in the WGEA report but not DESE’s figures.

This makes the WGEA data the best bet for analysing the early headcount employment effects of COVID. Unfortunately there is a significant problem with the total 2020 WGEA figures, as the University of Melbourne failed to report its casual numbers.

The non-casual WGEA numbers tell the same story as DESE’s figures – that in early 2020, with university managers blissfully ignorant of COVID-19 and Job-ready Graduates, university payrolls were showing the results of a hiring spree – up 3.7 per cent on the year before.

Excluding the U of M from both years, casual numbers were down 3,925 between 2019 and 2020, or 4.1 per cent (including the U of M, the total was 101,589 in 2019, 44.5 per cent of a total of 228,402). This is consistent with DESE’s estimated 2020 casual FTEs, down 3.7 per cent on 2019 figures. This is likely to under-state actual losses, since additional casuals had probably been hired prior to March 2020. Most indications between the 2019 WGEA snapshot and January 2020 were that the good times would continue.

Non-university higher education providers

Apart from Avondale in its pre-university college days, DESE’s staff statistics don’t include non-university higher education providers. Neither the ACNC or WGEA datasets provide a complete solution to this. The ACNC (obviously) does not include for-profits and WGEA does not include organisations employing fewer than 100 people.

I identified 22 NUHEPs on the ACNC list from late 2018 using a charitable purpose filter, less than the 41 a former Grattan Institute colleague found in 2018 cross-checking every TEQSA-registered provider against the ACNC database.

Using an industry filter, I identified 22 (same total as for ACNC) non-university organisations delivering higher education in first-half 2019 on the WGEA list, three of which were also on the ACNC list. I included some organisations that deliver higher education even though they are not registered with TEQSA. It is likely that 90-100 providers were either not identified using my search method, or did not qualify for either list, or had not complied with their reporting obligations (several still existing providers that reported to the WGEA in 2019 did not in 2020).

As many NUHEPs are small omitting them will not dramatically affect sector employment totals, but all the sources reported to date are sector under-counts.

Combining the late 2018 ACNC and first-half 2019 WGEA NUHEP numbers and removing the duplicates adds another 10,688 employees to the sector total, to 239,090.

The NUHEP sector – or at least those I have data on – is slightly less casualised than the universities at 38.4 per cent of all staff. The casual total goes to 105,696.

For a 2019 to 2020 comparison I have only 17 non-university providers with WGEA data for both years. They may not be representative of the non-university sector, but for what it is worth their total employment was down 4.7 per cent, including casuals down 13 per cent.

Fluctuating payroll numbers

An issue with these snapshot figures is that, especially for casuals, they don’t capture the full picture of annual employment. The ABS has started publishing data derived from payrolls, which gives us a deeper understanding of how employment changes during the year. Unfortunately they include higher education in ‘tertiary education’, with other data sources showing that higher education provides just over 70 per cent of all tertiary education jobs. And also unfortunately they publish an index without the underlying absolute numbers,.

Nevertheless, as the chart below shows, there are major fluctuations during the year, presumably largely driven by teaching casuals, with drops mid-year and more significantly over summer. This might suggest that the ACNC with a late in the year payroll snapshot date would under-count casuals, but its total is only slightly below the WGEA’s figures.

The chart also shows that with a base index of 100 on 14 March 2020, when the China travel ban was in place but the full travel ban was yet to be imposed, the index was at 96 on 13 March 2021. Applied to universities that would be a fall of about 9,000 jobs (including casuals) from the WGEA level as of late March 2020.

Numbers over a 12 month period

Using ATO data, in the TableBuilder version of Jobs in Australia the ABS has started publishing total higher education jobs during the financial year. This is based on the employer’s industry. If I understand the methodology correctly, with large multi-purpose organisations only the higher education part is included, when other data sources are counting everyone employed by universities and sometimes also their subsidiaries, whether or not the ‘activity unit’ they work in is related to higher education or not.

Due to delays in receiving and matching data Jobs in Australia is a lagging indicator, with the latest available being for the 2017-18 financial year. But this gives a much higher figure for annual employment contracts than the snapshot data, of 317,172. We need to keep in mind that the same person can hold more than one higher education job over a 12 month period (so that the number of employees is less than this) and that more than one person can hold the same position over a 12 month period (so that the number of ‘jobs’ is less than than this figure).

The number of ‘main jobs’ – defined as the job for which each person had the highest income for the year – was only two-thirds the level of total jobs, at 210,766. This definition means that people who joined or left higher education from or to another industry would be classified as having their main job in the other industry if it paid more in total, even if their higher education job was full-time and on-going. However, the large number of non-main jobs would be mainly the by-product of casual employment. For people with multiple casual higher education positions during the year only the one that paid the most would be counted as the main job, while no higher education position would be counted as a main job if the casual higher education worker earned more in another industry that financial year.

Conclusion

All these data sources on higher education jobs have problems – variously missing headcount categories (DESE), providers (DESE, ACNC, WGEA), total employment relationships during a year (DESE, ACNC, WGEA, ABS Payroll), distinct persons (all, though the snapshot figures probably only have minor double counts) or distinct jobs over a year (all).

Given the missing data for March 2020 no precise job total is possible, but it is plausible that there were around 250,000 positions in higher education organisations at that time, with a slightly smaller number of persons working in higher education due to multiple job holding.

It would be helpful if the ABS could release the absolute number of employees by industry in its payroll statistics, and split tertiary education into vocational and higher education.

The mid-March 2020 to 2021 ABS payroll data loss of 4 per cent is probably the best general guide right now to what has happened. If the 250,000 estimate is roughly right this would convert into approximately 10,000 fewer jobs across the industry. This does not mean that the NTEU or UA figures exaggerate job losses, since this estimate is a net figure that includes jobs made as well as jobs lost. The ABS payroll numbers show cycles of job creation and destruction.

Regardless of the exact number of jobs cut since 2020, the problems that began last year are far from over. The higher education workforce could easily shrink further during 2021.