Having a degree is having less of an impact on salaries. The financial return on a degree at age 26 fell eight percentage points in 20 years, with the rate of decline higher for students born more recently.
These are the tentative conclusions from new research by HESA and the Department of Economics at the University of Warwick, drawing on cohort studies following individuals born in 1970 (the British Cohort Study, likely to have graduated around 1991) and 1989-90 (the Next Steps datasets, likely to have graduated around 2011). The return on a degree – defined as the difference in hourly pay between graduates and non-graduates from the same cohort when controlled for skills and tenure – was 19 per cent for those born in 1970, and 11 per cent for those born in 1989-90.
Graduate returns – or sometimes the graduate premium – have become an increasingly live issue in policymaking, as successive governments use individual salary benefits to justify undergraduate loan levels. These new findings suggest that these justifications may no longer be as valid as they have been.
Checking their work
The HESA and Warwick team used data from the Labour Force Survey (LFS), which has been conducted quarterly since 1992, to construct similar cohorts to cross-check findings. This approach was very much a sense-check, less accurate than the cohort study data itself as there was less qualitative data on skills or other personal attributes to allow controls to apply.
The LFS method showed a decline in degree returns of 6 percentage points for men, and an alarming 25 percentage point decline for women. The cohort study approach showed similar (no statistically significant differences) declines in return for men and women, suggesting a static gender pay gap of around 9% over the 20 year period in question.
The overall finding of a reduction in the graduate return prompted the team to construct a series of birth cohorts using the LFS dataset, to try to identify when this decline may have happened. Again looking at hourly pay at age 26, the cohort born between 1980 and 1983 (with hourly pay therefore observed between 2006 and 2008) provided a baseline, with an average raw return on a degree at around 25 per cent. For those born four years later (between 1984 and 1987), there was a slight decrease – around one per cent. But for those born eight years later (between 1988 and 1991) the average raw return decreased by a further 8 per cent.
Making sense of it all
Clearly, and as the publication is at pains to point out, these are tentative findings. The relatively recent sharp decline in returns may be a temporary artifact – perhaps linked to wider labour market movements – or they may be the start of a more concerning longer trend.
Because of the availability of data the research looks at earnings at age 26, whereas we know from IFS research that graduate earnings grow steeply until age 30 – an issue we also face when using LEO data. There is more data coming from both birth cohorts, and the LFS will of course continue, so it is likely that later research will give firmer answers on these questions.
What could a policy response be?
This year’s Augar Report suggested a graduate premium for men at age 29 varies from minus 16 per cent to plus 58 per cent; with a median of 10 per cent, while the graduate premium for women varies from minus 11 per cent to plus 94 per cent with a median of 24 per cent. This was based on yet another methodology – the Institute for Fiscal Studies used the Longitudinal Educational Outcomes dataset but the findings are clearly in the ballpark of the later of the two HESA/Warwick measures.
The Browne Review of 2010 claimed that graduates would earn an extra £100,000 over their non-graduate peers, scaling down substantially an earlier forecast in the region of £400,000.
It’s clear that we have been dealing with a decline in the graduate premium for some time. As yet we have neither a convincing explanation for this trend – an increase in graduate numbers and a more difficult graduate job market have been suggesting – or any policy response. It’s clearly easier to assume that student loans are more than cancelled out by an earnings premium – this is already no longer the case for some provider/subject combinations and this disconnect is likely to become more prevalent.
There’ll be plenty of blame for universities, and plenty of castigation for the political aim of getting more young people onto degree courses. The behavior of employers is more likely to be the root cause, but it would be a brave minister indeed who points that out.