Don’t sleep on Jim’s exhaustive account of the various ways the Department for Education has calculated a graduate premium over the years – but do hold on to your hats as I reveal another part of the government has also been calculating future graduate earnings using a different methodology for a different reason.
To start with I pose a basic question – why do governments spend any money on education at all? It is, after all, a hugely expensive endeavor and one that requires a great deal of work without generating a direct economic benefit.
It is possible to imagine a world where, instead of starting school aged four or five, our young people immediately start paid work – obviously very simple tasks at first – and are trained up by employers as needed. Here, state expenditure on education would fall to zero, and (with more people in work) income from taxes would rise.
(Note to Richard Tice: this is not a good idea, don’t do this)
Just capital
The macroeconomic explanation for why we do otherwise is founded in the concept of “human capital”. To simplify: human capital is the capacity (in terms of knowledge, skills, health and many other aspects) that people have to generate economic returns. The more human capital a person has – the more they will earn. And the more people earn, the more they spend (thus driving the wider economy) and the more they pay in taxes.
The big education funding debates are, at heart, about using this additional human capital to cover the investment that the state has made in developing it. Funding flows from the treasury to education providers (who need to pay staff, run estates, solve complex car-parking problems and so forth) as an investment in the future earning potential of the people who benefit from this additional education.
If the government didn’t do this – employers would have to. And it is more efficient for the state to educate a large number of people than for the private sector to educate numerous smaller groups. The enhanced earning potential leads to a higher tax take, which is used by the government to offset the costs they incur in making it happen. All of which drives economic (and productivity) growth. That’s the theory, anyway.
The last year that the Office for National Statistics performed this calculation was in 2022 – it calculated the UK’s full human capital stock (measured by projected lifetime earnings of working age people) as £28.3 trillion: with average lifetime earnings per head of £566,000. These figures have grown by 1.2 per cent since 2011: and ONS attributes this to an increase in educational attainment.
Behind the curtain
How did they do it? The method relies on the Annual Population Survey (which is a larger annual version of the Labour Force Survey, so there is a larger sample size), and is used to identify (controlled by attainment, age, and gender) for employment rates, and salary rates. There’s also a built in allowance for survival rates (your likelihood of not being dead) but as mortuaries seldom record educational attainment it isn’t as accurate as we may like.
It assumes that an individual of a given age, gender and educational level will, in the next year, have the same labour income and other characteristics (employment rate, mortality rate, and so on) as those of a person who, in the current year, is one year older but has otherwise the same characteristics (like gender and educational level). It does, however, control for the likelihood that a person of a given age will increase their highest level of education in the following year. And there are a few other assumptions built in – growth in productivity is set at 2 per cent per year, the discount rate is set at the standard 3.5 per cent, and that there is no salary impact for additional qualifications after you have level 7 or above.
Here’s a plot of the “real full human capital” estimates for each year (which includes the impact of unemployment), split by highest qualification and age group (younger people have a greater lifetime earning potential at the point of estimation).
And here’s something I find even more interesting: if you look at each age group for a given year you will see that the earnings premium for higher education qualifications is more pronounced later in life.
This makes a kind of instinctual sense – skilled trades have a strong earning potential early in life, but the decline in health from long days of hard physical work eats into human capital in later years. But what it tells me is that in the narrower higher education funding conversation we pay way too much attention to early career earnings, and not enough to earning potential in later life.
(National) insurance premium
At every point each successive level of qualification holds a premium over the ones below it – again the differences are seen more prominently in mid– and late– career. And these premiums are an indicator of the investment that has been made in human capital by the state.
If you’ve read Jim’s piece, you’ll recall that DfE currently assumes the opposite – that after about a decade your earnings track record is a better indicator of future earnings than any qualification you might hold. This to me makes less sense for lower-level qualifications and careers involving physical labor because of those health-related human capital effects – it’s rather like assuming a high-earning professional football player will continue to earn at that level for the rest of their lives.
Before I leave human capital behind I also want to note that earnings potential isn’t a measure of human capital – it is an indicator. Human capital covers knowledge and skills (for which we can use qualification level as a proxy) but also covers things like social capital (who you know,) emotional capital (who you can work with or influence), and health (whether you can keep working or not). Salary is only a very simple way of measuring these wider benefits – and the reason people keep banging on about non-salary benefits to higher education is that they are important aspects of human capital that are very hard to quantify.
Death and taxes
So how about returns to the treasury? How much do we get back from people’s enhanced earning potential and does it cover the cost of doing so?
In the world of higher education, the gold standard for research like this is a 2020 paper by the Institute for Fiscal Studies – and we all recall that around 50 per cent of those who attend university will represent a net cost to the exchequer over their lifetime, even though 80 per cent will earn more over their lifetime than they would have done otherwise.
There are differences by subject of study – the creative arts is the poor performer everyone loves to leap on, but even more vocational studies such as nursing, social care, architecture, and (for women) the physical sciences are a net cost to the taxpayer. To be clear I’m not a big fan of the methodology here (I characterise it as an indication of what happened if people born in 1985 graduated in 2007 lived through to 2015 before travelling back in time to experience the decade 1975-1985 and then joining the 2019 labour force survey at age 40 and having earnings projected forward to retirement – all while paying back plan 2 loans).
The question that to me has never been asked is how this compares to the lifetime exchequer contribution (the amount paid to the state over a lifetime minus the amount “taken” in state services and allocations). The news isn’t good – whether or not someone has been to university the average cost to the state per UK resident person (spending minus payments) is around £400, with age as the determining factor.
You might be wondering why this data is from the Migration Advisory Council – it stems from a report that sets out in spectacular detail that migrants are net contributors to the UK while lifetime residents are not. We’ve only just got the data tables and they are a wonderful thing. This one is the contributions for an average individual of a UK resident in each age group in one year – 2022-23.
Lifetime guaruntee?
It looks even worse when you plot lifetime contributions. This plot of data from the same collection shows the mean and median future contributions (or cost) to the state for a person of each age in 2022. A UK resident child born in 2022, based on 2022 tax and spending plans) will, on average (mean) cost the state £102,000 (and the median, which is skewed considerably because state support is focused on the lower end of the income distribution, is £241,000.)
The higher figures for those who are older in 2022 are because we can ignore early spending in 2022. Discounting early-life healthcare and education spending, the average (mean) 18 year old in 2022 will contribute £167,000 to the state over their lifetime, while that skewing effect means that the median contribution is minus £7,000 – remember, stats fans, this means as many people will be below this number as above it.
So, if your metric for “quality” higher education is a net lifetime contribution, it is likely that the often cited average earnings premium of £100,000 and the wider benefits of enhanced human capital (less healthcare spend, greater civic contribution and so forth) would push you into this category as a graduate. And this would apply after loan repayments too: the value of the earnings “lost” would show up as a higher contribution “made” to the exchequer. On this calculus, loan repayments are only relevant to the lifestyle of the individual graduate – not the taxpayer.
The big policy problem facing any future government is the level of public spending per person – and the goal will be to at least break even on a per resident basis. As is usually the case when you follow the logic of neoclassical economics all the way through the solutions are firmly left of centre: a higher minimum wage, more effective lifelong healthcare, and a higher level of education, and enhanced lifelong learning, for all.
But how much of the outcome for any individual down to innate academic ability , and how much is down to how many things they learnt about in a classroom ? You don’t know the answer to that, and this data certainly doesn’t prove it. Nobody ever will know the answer as it’s impossible to work out. Yet you assume that the outcomes are down to how many years somebody spends in education and this is not a safe assumption (classic Correlation doesn’t prove Causation territory again I’m afraid)
Of course we can’t rely totally on innate academic ability and offer no free state classroom based education whatsoever and send everybody to work aged 5. But nor should we think that making the school leaving age ever higher (it is more or less effectively 21 now for anybody who wants a non-manual job with prospects) is a good idea for all. There must be a limit to the utility value of the amount of things we need our children / young adults to learn in the classroom before we push them out into the world of work where what matters arguably far more is getting actual experience of doing the jobs that society needs doing.
However, when we look at history, state provided education really only developed traction when the education levels (and so fighting capability) of conscripted WW1 British Soldiery measured up very poorly against the Germans. 100 years later, when we are talking about conscription, we are looking at unemployed graduates, “Plus ça change” as they say.
“But what it tells me is that in the narrower higher education funding conversation we pay way too much attention to early career earnings, and not enough to earning potential in later life”
This is an excellent point.