Skip to main content

Adding value to UK graduate labour market statistics: The creation of a non-financial composite measure of job quality - Section 5: Decent work

Section 5: Decent work for all?

In Irvine et al. (2018), the MJQWG pointed out the importance of examining job quality information by demographic characteristics ‘in order to understand inequalities in the labour market and target policy interventions where there is need’. Indeed, such work is necessary if the UK wishes to meet the United Nations Sustainable Development Goal of decent work for all. Among graduates, previous research by Zwysen and Longhi (2018) utilising the DLHE survey has established that once controlling for factors such as family background, university choices and work characteristics, there are minimal earnings differences by ethnicity six months after graduation. However, Clark et al. (2021) found that the gaps in ‘full earnings’ (which takes into account non-pecuniary aspects of the job) by ethnicity were greater than the differential when considering only monetary rewards. Given the results of Clark et al. (2021), we focus in this section on exploring whether disparities by ethnicity do emerge for graduates when focusing on our composite measure. 

In our sample, the mean score for the job design and nature of work variable was 4.00, with the standard deviation being 0.975. Using ordinary least squares, we estimate the equation below, where Yi is the composite measure score, while Ψi represents our set of controls and cover personal, study and employment characteristics. We use the white group as our reference category for ethnicity. Sex, age on entry, parental education (a proxy for family background/socioeconomic disadvantage), disability and qualifications held prior to beginning the course all form part of the personal attributes we control for. Study factors accounted for include mode/level of course, class of degree, as well as subject area and type of provider attended. Finally, employment-related variables utilised in the model are mode/type of contract, industry (Standard Industrial Classification 2007), occupation category (SOC 2020 major group), employer size and annual earnings. We also incorporated a marker indicating the extent to which the graduate had moved location for work and a categorical field highlighting whether their qualification was needed or advantageous in the role they were performing.   

Equation 1

Yi = α + β1indiani + β2pakistanii + β3bangladeshii + β4chinesei + β5black_africani + β6black_caribbeani + β7otheri + Ψi՛β8 + εi               

Table 1 provides our results and illustrates how the coefficients on the ethnicity dummies change as we successively add specific groups of covariates (those with an unknown ethnicity are not reported in the table below). Prior to the inclusion of any controls, we see that all ethnic minority groups report lower job design and nature of work scores, with the differences being statistically significant at the 1 per cent level. The addition of demographic characteristics tends to reduce the extent of the disparities we observe, though the coefficients for the Black African and Black Caribbean groups do not follow this trend. The reason this occurs is that these two ethnic groups tend to enter higher education at a later age (Connor et al. 2004), with older graduates reporting higher scores for our composite measure (see Table A2 in the appendix for further summary statistics on how our measure is associated with the independent variables utilised in our regression model). The inclusion of study variables leads to wider gaps emerging among all ethnicities within the Asian category (Indian, Pakistani, Bangladeshi and Chinese) and this seems to be driven by the variation in subject choices, with Asian students opting for fields that are associated with higher job design and nature of work scores. One of the key reasons why the differences become smaller for the Black African and Black Caribbean groups on inclusion of study characteristics is the class of degree variable. These two groups are less likely to be awarded a first or upper second class degree (Roberts and Bolton 2020), with higher attainment being associated with larger scores for our composite job quality indicator. Finally, the presence of employment factors generally leads to declining gaps by ethnicity (the Indian group is an exception) with this particularly being the case for the Pakistani and Bangladeshi graduates. That is, those from ethnic minority groups are more likely to be based in occupations and industries that score lower with regards to the design and nature of work measure.

Table 1: Ordinary Least Squares estimates of equation 1.

The dependent variable is the job design and nature of work score. Only ethnicity coefficients are reported. Standard errors are in parentheses. *** indicates significance at 1 per cent level.

  Controls included
Ethnicity None Personal characteristics Personal characteristics,
Study characteristics
Personal characteristics,
Study characteristics,
Job characteristics
Indian -0.0414*** -0.00540 -0.0645*** -0.0645***
(0.0102) (0.0100) (0.00957) (0.00785)
Pakistani -0.108*** -0.0445*** -0.107*** -0.0387***
(0.0122) (0.0119) (0.0114) (0.00936)
Bangladeshi -0.211*** -0.125*** -0.146*** -0.0802***
(0.0168) (0.0165) (0.0157) (0.0129)
Chinese -0.120*** -0.117*** -0.147*** -0.113***
(0.0209) (0.0204) (0.0194) (0.0159)
Black African -0.111*** -0.168*** -0.150*** -0.123***
(0.00925) (0.00910) (0.00875) (0.00720)
Black Caribbean -0.185*** -0.216*** -0.159*** -0.127***
(0.0169) (0.0165) (0.0158) (0.0129)
Other -0.128*** -0.117*** -0.115*** -0.0864***
(0.00771) (0.00754) (0.00720) (0.00591)
R-squared 0.00262 0.0484 0.138 0.424
Sample size 286,240 286,240 286,240 286,240

Within the UK, there has been a lot of debate in recent years as to whether ethnicity pay gap reporting should become mandatory for employers. Yet, no such attention has been given to whether we need to also monitor other aspects of job quality by ethnicity. We see that when it comes to the design and nature of work, graduates from all ethnic minorities report lower scores than their White counterparts. This is true even after controlling for a variety of personal, study and employment characteristics. Our results, along with those of Clark et al. (2021), thus illustrate the need for policymakers and employers to look beyond earnings and also assess inequalities in the non-monetary components of jobs undertaken by workers.

Next: Section 6: Concluding remarks

Tej Nathwani

Tej Nathwani

Principal Researcher (Economist)


Download research paper as PDF


See more research from HESA


Sign up for Research releases