Once in a lifetime: IFS data on earnings and contributions

Fully understanding the lifetime earnings premium for graduates has long been a grail quest for fans of higher education data.

The idea of graduates being better paid has underpinned decisions to levy and raise fees over the last twenty-five years – and has (as a part of wider arguments on economic impact) buttressed sector arguments on the student unit of resource since time immemorial.

The birth of the Longitudinal Educational Outcomes (LEO) data set allowed us to peer, for the first time, at salaries by provider and subject. Linking schools data, higher education data, and earnings data for individuals, LEO brought the idea of “low value” courses and providers into policymaking discourse. The capstone of this work was 2018s Institute for Fiscal Studies (IFS) report into HE attendee earnings at age 29 by provider and subject area – heavily cited in the Augar report and beyond.

IFS may well have been working on this more detailed look at these lifetime earnings – and lifetime exchequer income – by provider type and subject even as far back as the Augar report. It has been a long journey to produce what is an immense piece of work, and though I may not agree with some of the choices the team have made in crunching the data the work as a whole is satisfying and robust when taken on its own terms.

The headlines

IFS has estimated lifetime earnings, and lifetime exchequer receipts, for the entire working lives of those who have studied at HE level by subject area and institution type (Russell Group, post-92 other, other high tariff and other low tariff), split by sex..

• People who have studied in higher education will earn on average 20% more over their working life than those who did not
• After factoring in taxes and repayments, women’s earnings increase by £100k on average with HE experience and £130k for men over the course of their working life – after student loan payments and taxes are factored in.
• The economy gains £240k per man attending HE and £130k per woman on average over a lifetime.
• Eighty percent of those who attended university earned more than they would have done otherwise over a lifetime.
• Around fifty percent of those who attended university will represent a net cost to the exchequer over their lifetime.
• Variations by subject and provider group are more visible for men than women. This may be down to the greater likelihood that men will become very high earners, and the likelihood that women will take a family-related career break during this point.
• Creative arts courses appear to have a significant negative impact on private and exchequer returns, particularly for men. Economics, medicine and law courses have a positive impact on both counts, especially for men.

There’s no split by individual providers this time – there’s simply not enough data available to safely offer this.

The data

Estimated exchequer return and estimated net lifetime return

[Full screen]

This first visualisation plots estimated exchequer return (the amount paid to the Treasury over a lifetime via income tax, employer and employee national insurance contributions, and student finance repayments) and estimated net lifetime return (the amount earned minus tax and repayment). You can use the filter to examine and compare the impacts of provider type and subject area. Data for men is coloured blue, and data for women is coloured orange.

Estimated total return

[Full screen]

This second look offers estimated total return (broadly, estimated net lifetime return plus estimated exchequer return) by subject for men and women. I’ve shown percentiles (using shapes) in order to give you a taste of the variance in salary within each subject bucket.

Who are these people? (the caveats)

The people in question are not real in the fullest sense – they are a statistical chimera drawn from data from a number of different cohorts.

• Data from schools (the national pupil database) in England, HESA, and HMRC to age 30 came from the 1985-86 birth cohort – who primarily would have entered university in 2004 and graduated in 2007.
• To this we add – from HESA and HMRC data – the shape of the earnings growth of the 1975-76 birth cohort through to the 1984-5 birth cohort (data from age 30 to age 40, available for HE attendees only).
• Data on changes in earnings through to age 67 comes from the Labour Force Market Survey (a sample of 40,000 people each year) for those who attended HE and those who did not.
• Note, that for earlier years in particular, it is difficult to match HESA and HMRC data reliably.

And for income and repayment estimates IFS has assumed that the current (2019) system of fee and maintenance loans in England, and the current tax system, is in place. This data therefore does not represent the experiences of any single cohort.

You’ll note I’ve been trying very hard throughout not to use the word “graduate”. This data includes anyone who has ever attended HE to study a first degree – meaning those that failed to complete their course are also included. For those who switched between courses, the first subject studied is used for non-completers, and the final subject studied for completers. Joint and combined subject students are represented on a pro rata basis. I’m assured that this makes little overall difference to the figures, but it does concern me.

If the lifetime earnings gains seem smaller than you expect, the decision to discount later life earnings (using Treasury Green Book methodology) to calculate a “discounted present value” in 2019 terms has had a large impact. The calculation used can be broadly characterised as “how much money would you have to give someone now not to choose this course”. The discount rate is set at 3.5 per cent for the first 30 years and 3.0 per cent after that – other discounting measures are available. For this reason, dividing the overall figures by the number of years in an average working life to get an average annual premium is a particularly silly idea.

The top percentile for each subject group has been “windsorised” – with earnings allocated to the average of the 99th percentile. This process removes extreme outliers – for example Ed Sheeran’s earnings would not appear in the Creative Arts figures.

It is also important to remember that these figures make no attempt to account for hours worked – this data simply is not reliably available. It is likely that the earnings of women are lowered by this factor, and possible that some subjects may be affected too.

What might it mean?

Bearing in mind the above caveats, for me this data is most valuable in indicating proportional differences between subjects. The actual salary numbers are both interesting and tangible in presentational terms – but to use them to tell people how much more they might expect to earn if they go to university is not really helpful.

The Office for Students’ Nicola Dandridge says “It is crucial that prospective students have access to clear and factual information about what and where they might study, and data on earnings potential is an important part of that picture”. My own feeling, familiar to regular readers, is that there are too many confounding variables to share salary data like this in a meaningful way.

For example, I’d be nervous of using the provider group data – attendance at the Russell Group closely correlates with good A level results – both correlate with better life chances in early years. Rare counterfactual examples (a student with an access qualification and care experience, for example) would be rare in other ways too. There are controls in place, but I feel the difference in populations is too stark to account for.

If you’ve followed this line of analysis through Augar you will be aware of the governments’ keenness (as set out in the manifesto) to “tackle the problem of low quality courses”. Looking at this data, you could imagine that someone could claim creative arts courses – for men in particular – are “low quality”. Though a ministerial statement is initially emollient:

This research underlines that our university sector is world leading by setting out the impact higher education can have on someone’s life. When you add the unquantifiable experiences and friendships that come with that, it is no surprise our universities attract students from all over the world”

there is also language on “quality” and “value from investment” from DfE that should give us pause for thought.

Arts enrich all of our lives in ways that are notoriously hard to measure – there is a clear economic impact, but it is difficult to link this to any one person or set of skills. Artists, arguably, are notoriously underpaid – the basis of the gig economy, and likely to be taking on temporary jobs between engagements for their “main” vocation. Even so, the way arts subject are targeted – and the way the sector responds to that targeting – will define this debate for years to come.