On Wednesday the Institute for Fiscal Studies published the first report into graduate earnings based on the new dataset produced by merging HMRC tax data and information from the Student Loans Company on student borrowers. The report’s title sets out its range: “How English domiciled graduate earnings vary with gender, institution attended, subject and socio-economic background”.
Andrew McGettigan discussed the report’s findings and their implications with Jack Britton, Senior Research Economist at IFS, who authored the report along with Anna Vignoles, Neil Shephard and Lorraine Dearden.
What follows is based on three email exchanges that took place over two days at the time of publication.
AMcG: Were you surprised by any of the findings in this first report? or does it confirm what was previously acknowledged?
Jack Britton: Some things were unsurprising, others were surprising. The fact that graduates of medicine earn a lot, regardless of institution choice is not hugely surprising because the vast majority become doctors for whom pay is set by the government and exhibits relatively little variation. Similarly, the finding that LSE, Oxford and Cambridge graduates earn the most is not particularly surprising. On the other hand, the fact that students from wealthier backgrounds continue to have a large earnings advantage even after we allow for differences in their university experience is perhaps more surprising.
What would you say are the priorities for subsequent research and for policy?
In this paper we present a snapshot of the data showing graduate earnings for the 1999 cohort in their early 30’s. This does not give the whole picture and we would like to estimate lifetime earnings of graduates based on the early career trajectories we observe and on subsequent trajectories observed in survey data to give a lifetime view of the earnings advantage of being a graduate.
Your modelling and interpretation of the data sits within a human capital framework. Does this mean your ‘value add’ approach sees education as an investment that increases individual skills that are rewarded in an economy like ours by higher earnings?
There are many reasons for individuals go to university and many different types of return to any investment made in higher education, for example graduates may have greater job satisfaction. That said, our paper does indeed assume that higher education can increase skill levels, on average, and the higher earnings gained are one important part of the value of HE.
How could we control for motivation or selection bias? For example, that students who want to work in the City afterwards choose LSE. Should we see Economics at LSE doing a good job at securing desired posts for its graduates or is it that by replicating what LSE does one could replicate the returns you’ve found for a second set of graduates?
Although we try to allow for this, it is very difficult to rule out the role of selection on characteristics that we do not observe in the data. For this reason we do not claim that our results are causal. In other words, part of the high earnings of LSE graduates is due to the graduates themselves, and part of it is the institution effect. I therefore suggest it is unlikely that replicating exactly LSE’s teaching, facilities and location would give the same results for all other students, since, for example, other students may be more or less oriented towards some of the higher paying jobs that many LSE graduates do.
Medicine is clearly a special case – what happens to your institutional comparisons if medicine is excluded?
Because medicine is a relatively small subject group, it might not make a huge difference to the overall results. However at institutions where a large fraction of the students are medical students, this is likely to have an impact. We haven’t tried this, but I agree it would be interesting to look at.
Graduates of Creative Arts degrees came out worst in the report’s findings: they “had the lowest earnings, and indeed earned no more on average than non-graduates”.
Creative Arts attracts students with a different set of motivations and different early career paths. For example, part-time paid work to support creative practice while attempting to establish a professional practice is normal. How should we understand those differences in relation to your findings?
And at the same time, could I also ask about part-time working? I take it you can’t capture paid hours worked from the HMRC data, but I am interested more broadly in how you see certain kinds of early career structure fitting into your analysis.
In your framework, how should one interpret the choice to reduce hours worked so as to free time for other activity? In human capital theory, ‘psychic income’ is one return on education investment. If people are able to support themselves by working fewer hours after a degree isn’t that a benefit from their education?
We need to be clear here. We are just looking at earnings, i.e. just one aspect of the many potential benefits of higher education. So creative arts may well yield high levels of psychic income but we were not able to look at that though we are clear that non monetary returns to a degree are important. On the specific issue of hours, earning the same through reduced hours would definitely count as a return on a private investment, but because we can’t observe hours it is difficult to comment on this further.
We have the population of borrowers, so that includes everyone, regardless of their hours (and including those with zero hours/earnings). However, we only see total earned income in a given year, so are unable to look at hours or even the distribution of earnings through the year.
It was beyond the scope of this paper to consider lifetime earnings, and we just provide a snapshot 10 years out. In future work we plan to simulate earnings through the lifecycle, though it will be difficult to incorporate later career jumps in earnings that are specific to creative arts type subjects without good quantitative evidence of these types of patterns.
Am I right to assume that specialist creative arts & design institutions are included in the database? If so, was any consideration given to grouping them separately, particularly since these subjects would not normally select on UCAS tariffs but on portfolio and performance?
Yes some specialist institutions are in the data. Some will have tariff scores, some will not. Note also the data are provided on the basis of the classification of subjects used in the JACS classification system so we had limited ability to regroup specific subjects and as you know Creative Arts covers a wide range of subjects including subjects taught in non-specialist institutions.
Given the focus on the low earners registered by creative arts graduates, were any specialist institutions contacted regarding permission to publish their data?
As I think you are aware, we had to ask permission from each university to be named under the data access rules. We started with the Russell Group since we thought they would be most likely to agree to being named. Clearly we had to explain the research to these institutions to gain permission and we are bound by the limits of that. We did not contact specialist institutions nor any other institutions, though we think this is a good suggestion to pursue for the next phase of the work.
AMcG: At the March Budget, the Chancellor reaffirmed the government’s commitment to publishing earnings information by course and institution. Do you have any reservations about publishing the raw data in this way?
Jack Britton: For the purposes of really understanding the value added by subjects and institutions the raw data is not very helpful since it does not allow us to compare like with like which is why in our paper we take a modelling approach that tries to account for differences in student intake, for example. However, it should be said that lots of information at the course and institution level is already available.
What implications does your work have for the proposed incorporation of earnings data into the TEF? Should this have a value add dimension? Should it be by subject?
Including information on employability and earnings in the TEF may be useful as part of a basket of measures. It would indeed be important that any such measure took a value added approach to allow for differences in student intake and an average across an institution which included a different mix of degree subjects would not be comparable with another institution with a different mix of subjects so data at institutional level could be problematic in that regard.
Would it be better to fund specialist creative arts and design institutions differently given that their graduates are unlikely to repay much of their loans?
We would not want to comment specifically on funding of specialist institutions since we did not look at that in this work. More generally it is important to be transparent about which subjects and degrees are being subsidised to a greater extent and given better information on this.