The returns to apprenticeship training

Abstract

This paper uses recent data from the UK Labour Force Survey to estimate the wage gains that individuals make on average if they complete an apprenticeship programme. The results suggest gains of around 5–7% for men, but no benefit for women. Further analysis extends the results by considering the returns by age group, by qualification obtained, by highest prior qualification and by industrial sector. A key finding emerging from this further analysis is the importance of acquiring level 3 qualifications with the apprenticeship.

Acknowledgements

The Labour Force Survey data used in this paper were made available by the ESRC Data Archive at the University of Esssex. The research was funded by the Esmée Fairbairn Foundation, as part of a programme of research entitled ‘Skills for All’. I am grateful to Hilary Steedman for helpful comments.

Notes

1. The body of evidence accumulated on this issue is most closely associated with the National Institute for Economic and Social Research, see e.g. Prais, 1995; Oulton, 1996; Mason, 2000.

2. See e.g. Gospel (1997), Steedman (2001a, 2001b) and Steedman et al. (1998).

3. Fuller and Unwin (2001) offer a description of sectoral differences in apprenticeship quality.

4. See, however, useful case‐study evidence on employers’ costs in a recent paper by Hogarth and Hasluck (2003).

5. Attempts to calculate what trainees would have earned had they not been in training have been made by Payne (2001), using data from the Youth Cohort Study to compare the wages of those involved in AMAs and ‘other GSTs’ to those in a full‐time job.

6. See e.g. Payne (1995) for firsthand evidence.

7. Prior to 1996, the LFS asked respondents simply to list their highest three qualifications. Since the analysis that follows benefits from knowing all qualifications held by respondents, this effectively determines the starting date to be 1996.

8. There are also significant numbers of respondents to the LFS who indicate that they hold City and Guilds qualifications, but are unsure as to which level. They were therefore included as a separate category in the analysis.

9. The difference between diplomas and certificates at each level is that the former are studied full‐time at an institute of further or higher education, while the latter are taken by part‐time candidates usually in employment. In the context of apprenticeship, it is therefore likely that the certificates will be awarded.

10. NVQs were originally intended to certify the possession of relevant skills in a workplace setting, while GNVQs were less occupation‐specific and more likely to be taught outside work. The distinction has blurred somewhat with the provision of full‐time NVQ courses in FE colleges.

11. In addition, separate equations are estimated for men and women, since the available literature suggests that different wage‐generating mechanisms seem to be at work for the genders.

12. See Keep et al. (2002) and Machin and Vignoles (2001) for a fuller discussion of such issues in the context of employer‐provided training.

13. See Dearden et al. (2002) for estimates of returns to a fuller list of qualifications than presented here, including academic qualifications.

14. Since all qualifications are included in the estimated equations, rather than simply highest qualification held, the coefficients represent the additional returns specifically to acquiring each qualification, and are cumulative across any combination of qualifications.

15. In fact, the estimated returns to these qualifications are always negative. This is likely to be a consequence of unobserved characteristics biasing the coefficients, with the individuals who acquire such qualifications being of lower unobserved ability, and so earning less in the labour market for this reason. The coefficients should not be interpreted as implying that wages will actually fall following the acquisition of NVQ qualifications at levels 1 or 2. All that can be said here is that there is no evidence in favour of these qualifications having any positive impact on wages at all.

16. There is apparently a 12% return to an ONC/OND qualification for women in 1996, although this seems to be out of line with the results in other years, and could be a statistical artefact.

17. There is some evidence, not reported in Table III, that higher‐level RSA qualifications, typically secretarial, have a positive return in some years for women RSA qualifications were not reported in the tables, however, because they are not usually found in conjunction with apprenticeship.

18. It is true that the interaction coefficients for the lower‐level City and Guilds qualifications (Foundation level and ‘don’t know which level’) are positive and statistically significant. However, the base coefficients for these two qualifications are both negative, meaning that the net effect if an apprentice were to acquire them during his apprenticeship would be essentially zero.

19. It is the case that some apprenticeships in engineering can result in the award of an HNC/HND.

20. The estimates of NVQ/GNVQ returns at levels 4 and 5 are now well defined due to the larger sample size in the pooled data set.

21. See Dearden et al. (2002).

22. See Dearden et al. (2002).

23. Because of these low numbers of A‐levels students going on to apprenticeship, these individuals are not considered in the remainder of this section.

24. An alternative interpretation of these results across prior qualification groups is that they are affected differently by the selection bias, described above. In particular, among the group with no school qualifications at all, those who nevertheless acquire an apprenticeship, might be expected to be of above‐average (unobserved) natural ability, while in, say, the group with five good GCSEs, it is not clear whether natural ability will be greater or smaller among those who complete an apprenticeship. Thus an upward bias to the estimated returns caused by unobserved ability may be more likely to emerge among the group with low‐level prior qualifications. There is, unfortunately, no way to investigate this possibility further with the current data set, but it should be borne in mind.

25. Because of small cell sizes when looking within industries, the small numbers with NVQ/GNVQs at levels 4 and 5 are added to those at level 3 for the sectoral analysis.

26. This figure is for the under‐25s in the data set, to ensure that only recent apprentices are being considered.