A whole range of people – the Department for Education, the Institute for Fiscal Studies, (briefly) the Office for Students have been torturing LEO data in the hope of finding out which are the “good” universities and courses.

LEO is, of course, Longitudinal Educational Outcomes data – the fruits of a sordid co-mingling of tax and educational datasets pioneered by David Willetts. Although it is an interesting dataset in its own right it has a number of issues that present themselves most clearly when you use it to determine which course and provider will land an entrant the most impressive salary a few years on from graduation.

I’ve explained the problems with LEO for doing this many times before – if you want a recap why not start with the last time I plotted it? But now the newly Katharine Birbalsingh-free Social Mobility Commission has had a stab, in the form of what is best described as a literature review. And the key phrase is “value add”.

On average, of course, a person will be better off for having taken a higher education course than a further education course. That’s inarguable – even if you take personal characteristics into account. But there is some wiggle room when you get into individual subject areas and providers, so (for instance) IFS found out that men who studied nursing or creative arts would (on average) have been financially better off over their lifetime if they hadn’t bothered.

But as SMC notes, this isn’t quite the whole story. Certain providers – and certain courses – are more likely to attract students from disadvantaged backgrounds: students who would be expected to earn less well than their peers in any setting. Disadvantaged students are more likely to have lower prior attainment, and more likely to have caring responsibilities to give two of many common examples.

But even after controlling for personal characteristics, attending (any) HE is associated with 19 per cent higher earnings for men and 24 per cent higher earnings for women – with the figures less clear for further education , although the higher the level of education the greater the benefit. There’s also a benefit for apprenticeships above other vocational courses of a similar level.

The caveat here is that we assume what SMC calls “unobservable factors” – stuff like personal motivation, personality – are not driving earnings differences. Which is very clearly not the case.

So when we hear stuff around more selective providers or courses leading to better salary returns – and thus better social mobility – there’s a whole causal chain that we need to unpick. If I’m honest this isn’t the report that does that – but it is at least encouraging that SMC are engaging properly with the question.