LEO is, as I’ve often said, a great data set for social scientists, historians and economists – in that it allows them a warts-and-all view of the graduate labour market over the past decade. It is of vanishingly small use to anyone looking to understand anything about the quality of teaching in a given subject or institution.
This time round we have a postgraduate dataset – allowing us (when we combine it with the undergraduate data from earlier this year) to make an estimate of the difference in median earnings and employment status between graduates, postgraduates and doctoral graduates.
But before we get into that, here’s the basic UK PG LEO data as an interactive for you to play with.
The controls allow you to select the level of study, choose one of the two available tax years, look at gender splits and choose your subject area.
One notable effect you can see here is lower salaries for Education and Law at Level 8 for those who qualified 10 years ago than 5 years ago. These are the only two instances that buck the general doctoral trend of those with more post-doctoral work earning more than those with less. Those with doctorates in these two disciplines are overwhelmingly likely to work in academia – I’ll come back to this.
Comparing apples, pears, and other orchard fruit
If we combine this dataset with the recent undergraduate LEO release, we can start having a little more fun. This next sheet has three tabs:
Median salary over time looks at levels 6 (undergraduate degree), 7 (masters degree), and 8 (doctoral degree) for any given subject. In nearly every case at every sample point you see level 8 earning more than level 7, which in turn earns more than level 6. But the exceptions (looking here at the 2015-16 tax year) are interesting:
- In Medicine a masters’ level qualification has a negative impact on earnings after 3, 5 and 10 years.
- In English, a masters’ level qualification makes very little difference over an undergraduate qualification after 3 and 5 years, and is an active detriment after 10.
- In Economics, a masters’ level qualification is better for earnings than a Phd.
It really sucks to be a PhD in the 21st century
Sustained employment over time is where things get interesting. In nearly every subject, a person with an undergraduate degree is more likely to be in sustained employment and/or further study than someone with a masters’ degree, with a PhD doing even worse. Smaller subject show more volatility – for example Psychology is the opposite way round. And looking at all subjects shows a split between taught and research postgraduate, with the latter ending up as a detriment.
The main text suggests that this may be linked to less data captured due to more PhDs working overseas. Our alternate theory as to the reason why is exquisitely painful – the higher education sector is a really terrible employer. Those with research degrees are more likely to attempt to pursue an academic career, spending a decade or more chasing post-doctoral research contracts or sessional teaching.
The definition of “sustained employment” is fairly weak – “one day for five out of the six months between October and March of the tax year in question or if they had a self-employment record in that tax year and earnings from partnership or sole-trader enterprises of over £0 (profit from self-employment)”. Literally five paid sickness cover tutorials could put you over the bar here, or a single shift as a Deliveroo courier.
And further study can be any “valid higher education study record at any UK HEI on the HESA database in the relevant tax year. The further study does not have to be at postgraduate level to be counted”. The first couple of sessions of a mandatory teaching qualification would count here.
But even with these low expectations one in five PhDs are unemployed three years or more after qualifying.
Another attempt at winning Sam’s app competition
The final chart plots median salaries (plus lower and upper quartiles) for each subject area, filterable by years after graduation, tax year and gender. If you’ve graduated since 2003 you can look up your approximate cohort and subject to find out if your earnings would be better if you’d done that PhD or masters’ – or indeed whether that PhD or masters’ was worth it in terms of salary.
It’s not really any use for predicting future returns, but as an engine for generating baser human emotions like regret, pride, or envy it is pretty much unsurpassed. Feel the raw joy of being in the upper quartile, or the sheer pain of being in the lower. Laugh cruelly at people who spent five years doing a PhD and still likely earn less than you.
Then look in the mirror and have a long hard think about the kind of person you are and the kind of world you want to live in.
The experiment continues
For all the noises off about using LEO to make judgements (and funding decisions) on the value of particular subjects. We’re in our second year of experimental statistics, and there’s still not even a consensus on what the data means or what can be inferred from it. Further research is – as they say – needed. Let’s hope they employ someone with a PhD on a decent contract to do it.