graduation day

Since the early 1960s, with developments in the field of human capital research, analysis of the returns to education has established robust evidence of a strong positive association between earnings and years of schooling or level of qualification attained. But there has been little analysis of how returns vary according to the level of academic performance – for example, what US universities measure as students’ ‘grade point average’ – conditional on the level of qualification.

This is particularly surprising given that employers often recruit at specific educational levels and, when ranking candidates, they are likely to consider grades or marks achieved. For example, according to a 2010 report from the Association of Graduate Recruiters, 78 per cent of employers filter out applicants who have not achieved at least an upper second.

Our research investigates whether there is a premium associated with the level of educational achievement at university for graduates in the UK, as measured by the class of degree awarded: first, upper second, lower second or third.

Graduate earnings might vary with academic performance at university because employers treat performance as a signal of potential productivity. In that case, we might view a high premium as reflecting employers’ lack of information about workers’ productivity at the point of recruitment, and expect the premium to diminish with workers’ tenure, as employers acquire information about their productivity.

Graduate earnings might also vary with academic performance at university because degree class is a measure of human capital acquired. Under that approach, we would interpret a high premium as reflecting greater acquisition of human capital by the graduate. Under either approach, evidence of a high premium associated with the level of academic performance is likely to be interpreted as indicating substantial variation in post- university outcomes by degree class.

Our study estimates the returns to degrees by degree class awarded by analysing data from the British Cohort Study (BCS70) for a single cohort of graduates born in a particular week in April 1970 and graduating in 1991. We obtain an estimate of a wage premium of 7-8 per cent for a good degree (a first or upper second) relative to a lower degree (a lower second or third) at the ages of 30 and 38. We view the estimated premium to be large when we consider that our estimate of the premium for a lower degree relative to A-levels is 11 per cent at age 30.

The implication of this result is that there is a large dispersion around the average return to a degree according to class of degree awarded. Such evidence of substantial variation in graduate returns by academic performance creates a public policy concern if the perception that investment in higher education is risky acts as a deterrent to participation. In particular, young people from backgrounds where participation in higher education has traditionally been low might be less confident of their capacity to perform well.

We supplement evidence from the BCS70 with analysis of other datasets, such as the Labour Force Survey (LFS), the Higher Education Statistics Agency’s survey of destinations of leavers from higher education, and graduate cohort studies. Using these datasets for graduates born in or around 1970, we obtain estimates for the good degree premium that are very similar to the BCS70 estimates.

We conclude that there is robust evidence of a significant wage premium for a good class over a lower class of degree for individuals born in or around 1970 and graduating in 1991. At least for graduates more than five years out of university, our estimates of the wage premium all lie in the range of 6.8-8.8 per cent. Evidence from all data we employ suggests that between the ages of 30 and 40, the wage premium for a good degree is essentially constant.

We also examine variations over time in degree class premia. Using information from the LFS for the period 2005-13, we produce estimates for birth cohorts ranging from 1973/74 to 1981/82. This enables us to make comparisons of the premium for a good degree across birth cohorts at specific ages, covering the period of UK higher education expansion in an attempt to identify any trends in degree class premia across cohorts. We group individuals by birth cohorts, producing estimates of returns to degree class for individuals at ages 28-31, 30- 33 and 32-35.

Our estimates of the premium for a good degree tend to cluster in the region 9-11 per cent, but they do not reveal robust evidence of any clear upward or downward trend. We note, however, that this analysis is based on comparisons across cohorts through a limited observation window and with relatively small sample sizes.

We address this issue by exploiting administrative data on full populations of university graduates, matched to survey data on their first destinations after university. We find that the early career earnings premium associated with a good degree was very modest (less than 3 per cent) until the index of participation in higher education began to rise markedly across the cohorts graduating between 1991 and 1998, by which time the premium had increased to more than 6 per cent.

We also find evidence that much of the increase in the premium for a good degree after 1990 was realised by 1993 as the index of higher education participation rose from 15 per cent towards 20 per cent of the age cohort. These results are consistent with the idea that as more young people obtain degrees, the premium for graduating with a good class of degree increases.

We explain the absence of clear evidence of further increases in the good degree premium by the observation that the proportion of students awarded good degrees grew markedly between 1993 and 1998, when the expansion of higher education tailed off. We also find that the premium associated with the award of a first class degree (relative to an upper second) grew significantly over the period of expansion.

Due to a lack of data, we cannot identify whether the estimated premium for a good degree arises from signalling or because degree class is a proxy for underlying ability and other dimensions of human capital. If degree class acts as a crude sorting mechanism for graduate employers, then this might be a further justification for the current trend in the UK away from the traditional system based on degree classifications and towards the issuing of detailed transcripts and grade point averages.

But if the continuing marketisation of higher education in the UK leads to further grade inflation through the awarding of a higher proportion of good degrees, then this is likely to reduce the premium.



Robin NaylorRobin Naylor is Professor of Economics at the University of Warwick. His research interests are in the area of labour economics, including economics of higher education; intergenerational mobility; labour market outcomes and product market influences; economic integration, trade and wages; and social customs in the labour market, trade union membership and bargaining outcomes.


Jeremy SmithJeremy Smith is a Professor of Economics, Deputy Head of the Department of Economics at the University of Warwick and Chair of the Social Science IT Committee. His research interests are in time series econometrics; the theory of forecasting and forecast evaluation criteria; non-linear models; and higher education outcomes and performance.


Shqiponja TelhajShqiponja Telhaj is Senior Lecturer in Economics at the University of Sussex and a research associate in CEP’s education and skills programme. Her research interests are in applied microeconomics; economics of education; and labour economics.