What could a registration refusal for a college mean for higher education?

So Newham College becomes the latest casualty of the Office for Students provider registration process.

It gets a refusal on the basis of Condition B3 – which has also featured in the refusals of Waltham Forest, ABI College and Bloomsbury Institute:

The OfS takes the view that the provider’s performance in relation to continuation rate data (evidencing the number of students progressing from their first to their second year of study) shows that the provider has failed to demonstrate that it delivers successful outcomes for all of its higher education students, which are recognised and valued by employers and/or enable further study for all of its students.

We know that there’s a threshold that triggers a B3 condition of approval, and so far as well as headline refusals, two universities and sixteen colleges have had enhanced monitoring conditions for not meeting it. FECs are especially vulnerable here – they have enough data banked with HESA and HEFCE to allow OfS to make a “sound” B3 judgement (if perhaps not a statistically meaningful one) but a slice of the new providers on the register haven’t. Wait until their second student number return, and plenty more could emerge.

Let’s set aside for a minute the questions surrounding DfE’s previous funding of these providers, and what happens now for a college which at the time of writing is “open for enrolment NOW” (their capitals).

And we’ll also ignore that muddier question of continuation rates being something that OfS should use – especially when students are from more “challenging” backgrounds. That debate goes something like “they’re bound to have worse rates given X Y Z” and indeed the Association of Colleges has called on OfS to “consider the context of colleges and the role they play in communities”, but then others retort with “they obviously couldn’t cope with the course so why did you recruit them/not support them better” – and you can see why the latter side wins.

So what?

Let’s look instead at what this all might mean. One of the grumbles around the sector has been that the figure that OfS requires on, for example, continuation isn’t revealed. We hear that it might be “contextual” but that could mean the context of the subject, the provider, the year, the intake or who you happen to have from the registration team on the phone. In any event it is very odd that a regulator regulates on hard numbers but can’t offer a clear line on which numbers are and are not acceptable.

In time presumably that number (or method) will reveal itself, or will be revealed by multiple refusals. It’s what happens next that gets tasty.

It’s the economies, stupid

Large, mainstream universities have economies of scale, which famously for example allow for cross subsidy between courses that are expensive to teach and those cheaper to teach. The truth is that that scale also allows for cross subsidy between provision when it comes to performance data. For example – a large, mainstream university holding its region’s nursing contract (at least until recently) will have done better on POLAR and Mature, balancing out bits of the uni that don’t. But given the historically poor continuation rates for nursing, other bits of the university will need to be better at retention to balance things out. The bigger you are, the easier it is to play those figures in a broad portfolio.

We’ve argued before that this is a problem in all sorts of ways – not least because access to the “professions” is measured through Russell Group access in general as a proxy – but what if, across the Russell Group, law and medicine were not getting better at widening participation but english and social science are? How would we even know?

Playing those numbers is not so easy if you’re small and local, giving life chances to people to people who don’t want the “boarding school” model. So let’s run a thought experiment.

Work with me

First, let’s take the overall size of a single provider refused on the basis of B3 and specifically its continuation rate between Y1 and Y2. According to TEF workbooks, Waltham Forest was on about 150 students – reasonable for a hyperlocal FE provider. And according to its TEF4 workbook, its continuation rate was 65.8%.

Now let’s see if we can find a single course – or at least a single aggregated set of courses – on unistats that is as big as Waltham Forest, but has a worse continuation rate. We can can confirm there are such courses (or at least aggregated unistats clusters of courses). And they often exist in very prestigious providers for whom great claims are made about teaching quality.

As a special treat – here’s what the July Unistats data (the dataset that students in clearing would have used) told prospective students about non-continuation at course level:

[Full screen]

Note the definition of continuation unistats uses, and the aggregation level included in the tooltip. It’s also worth bearing in mind that Unistats data is (usually, but not always) rounded to the nearest five per cent.

This is a problem for all sorts of reasons. If (and it’s a big if) we end up with Gold, Silver and Bronze medals at both subject (TEF) level and institution level, it’s already preposterous that in a few short years a given department doing a sample lecture at an Open Day might have TEF Bronze and TEF Silver provision in a TEF Gold university. We can guess which banner will go up in the school reception.

But away from public information and aggregation basket concerns, this surely represents a regulation issue. That Y1 to Y2 metric is in the TEF and has been in Subject Level TEF pilots (and is of course a projection not a measure, as it is 18 months out of date). Remember that line in the Gavin Williamson letter about the TEF project being “seamlessly integrated” into OfS’ regulation more broadly, and lifting the “baseline” to ensure providers deliver successful outcomes for “all students”?

It’s obviously not the case that he literally means “all students” – in that event a provider would have to have a Y1 to Y2 continuation rate of 100%. But it would surely be odd if a provider with 150 students is refused registration altogether on the basis of that figure, only for another larger provider to be running whole courses with more students on them than 150 with worse figures.

There are other potential unfairnesses. If you provide HE in an FE provider, it’s the data for regulated HE courses that will be looked at. What if the data is below a given threshold but would be fine the context of your overall provision? Are we saying that a large HE-only provider can play the averages but a large FE provider with some HE can’t? Even worse, what if the data for your regulated HE provision is below the line but the rest of your HE is significantly above?

OfS’ next move

There’s only two things OfS could do about these manifest statistical unfairnesses. On the one hand, you could allow the micro-providers on, by judging a bunch of them as a basket or implementing ever more complex benchmarking. Both seem unlikely.

Alternatively, as a regulator you “notice” the continuation rates on those courses and you act. First there’s the question of aggregation – do you use Subject TEF type aggregation, or some other basket? Would you treat franchised provision of a single course as a block or franchise provider by franchise provider? Can you prevent crafty moving of students between subject categories? How do you benchmark?

Then there’s the question of what you do. We are assuming that OfS doesn’t have the powers (or desire) to deregister an entire provider on the basis of one course’s continuation rates. But we suspect it may have to issue conditions of registration and fines on that basis – especially if they plump for the subject-TEF subject categories as the relevant statistical baskets.

Once you’re there – and if you follow the logic, they probably should be there – you end up having to implement subject level (or least subject-TEF level) access and participation regulation – lest your approach in that space looks out of kilter with your provider registration process and your TEF process.

So perhaps most importantly, there’s then then question of whether OfS can do all that – legally. Most of us have got a vague idea that the Higher Education and Research act prevents OfS from regulating at subject or course level. But we’re not so sure.

Legal eagles

First of all, in performing its functions, OfS must “have regard to the institutional autonomy” of providers which is defined as determining course content and delivery; criteria for the selection, appointment and dismissal of academic staff and criteria for the admission of students. So far, no bar.

Next, OfS must have regard to guidance given to it by the Secretary of State (and that guidance must have regard to the need to protect institutional autonomy). The guidance may be framed by reference to particular courses of study but it must not relate to “particular parts of courses of study”, “the content of such courses”, “the manner in which they are taught, supervised or assessed”, “the criteria for the selection, appointment or dismissal of academic staff, or “the criteria for the admission of students, or how they are applied”. Again, nothing there.

It’s the next bit people (mis)remember. “Guidance framed by reference to a particular course of study must not guide the OfS to perform a function in a way which prohibits or requires the provision of a particular course of study”. If OfS was to say “all of your courses” or “all of your subject TEF clusters” have to hit a particular continuation rate, it would neither be prohibiting nor requiring. It would just be saying – make those courses or subjects better.

Fascinatingly, public debate on Augar and VFM has focussed on LEO data – but non-continuation is a big VFM issue too. And on P102 of Auguar, it’s right there – on so-called “low value” courses he says OfS should monitor progress in terms of “non-continuation rates on lower value courses.”

And all of that… means this. For large university providers, the days of economies of scale enabling you to cross subsidise your outcomes figures too are probably numbered. And depending on who you are, that’s either excellent or absolutely disastrous news.