The REF outcomes are used in three different ways: to demonstrate the quality of UK research, to burnish the reputation of specific universities, and in England, to distribute £1.5bn via Research England, and specifically £1bn of quality-related (QR) income.
Elsewhere on the site, James Coe provides a stout defence of the principles of QR. Here I’m going to probe the levers and pulleys that Research England can operate to affect QR allocations.
Where it came from
It’s probably helpful to describe how we’ve got to where we are, and the current QR methodology.
Before RAE08, assessment outcomes were less finely grained, and a unit of assessment (UoA) was awarded one of seven possible ratings. QR funding was only awarded to the 3 top rated grades (4, 5 and 5*). So there was much modelling of submissions to determine the likely cliff edges of funding – if you submitted one too many staff, and were awarded 3a, rather than a 4, then that would be seven years of zero QR income for the UoA.
Each of the ratings had a different weighting attached to them in the QR formula. Over time the relative weightings between 4 and 5/5* were also amended depending on the amount of QR that was available for distribution. The variances weren’t large, but it had the effect of concentrating QR slightly further into the highest rated UoAs.
RAE01 quality rating Funding weighting (in 2008-09
With the change in RAE08 to a quality profile, HEFCE had to allocate QR in a completely new way. Profiles meant that the funding cliff edges of the old model disappeared. A unit that had been a 3a unfunded UoA may now have had a small percentage of 2*, 3* or 4* research. The instruction to HEFCE from government was to “continue to selectively fund …excellent research wherever it is found”.
|Quality level||Weighting 2010-11||Weighting 2011-12||Weighting 2012-13||Weighting 2015-16 to present|
As the table above shows, the threshold for the minimum quality of research to be funded has changed over time. With 2* research only making a brief appearance in the funding formula the model and the parameters have been largely unchanged by HEFCE, and then Research England. And what they haven’t done is to tweak the relativities based on the funding available as they did prior to RAE2008.
The volume measure of research has barely changed over time – the FTE of staff submitted to the RAE or REF. There are also subject weightings to take account of the different costs associated with research: 1.6 for STEM, 1.3 for mixed subjects, and 1 for “classroom” based subjects. These subject weights have been a near constant for over twenty years.
With the changes in quality weightings after each exercise, there has also been an increasing diversity of institutions receiving QR. The share going to the Russell Group has been slowly decreasing after each REF, even with the weightings shifting towards 4*.
Share of QR 2007-08 2014-15 2021-22 2022-23
Russell group 71% 69% 68% 64%
Other providers 29% 31% 32% 36%
(The historic figures include the members of the Russell Group as constituted now)
If no changes are made by Research England after REF21, then the Russell Group could end up receiving their lowest ever proportion of QR – modelling suggests this could be just 64 per cent of the QR pot.
QR allocations could be seen as the antithesis of levelling up. Funding is concentrated into a handful of established universities. It is even one of the last bastions of London weighting (£34m is allocated to London institutions in addition to their QR allocation). Research England will need to determine what it means to continue funding excellent research wherever it is found.
The levers available to Research England appear to be quite limited. They might also receive more specific direction from government about whether there should be more or less selectivity. Moving to funding only 4* research would make the allocations more selective, with the Russell Group returning to ~68% of the available pot. Removing the London weighting would have a similar impact in terms of selectivity, although it would effectively be redistributed within the Russell Group!
With recent changes to subject weightings in the teaching funding model, there could be parallel guidance to Research England. Government could direct QR allocations to “high cost strategically important subjects” and away from “non-high cost, non-strategically important subjects”. The current allocations put ~64 per cent of QR into STEM subjects. Modelling of the REF2021 results means that this could reduce to ~61 per cent for STEM. However, it would require very radical changes to move this dial – reducing the weightings for art, music and dance might only release 1 per cent of QR to be redistributed.
There are no easy answers in the QR conundrum – every possible change will have winners and losers which will have an ongoing impact until the next REF.