CLASS ORIGINS, EDUCATION AND OCCUPATIONAL ATTAINMENT IN BRITAIN

ABSTRACT

Studies of intergenerational class mobility and of intragenerational occupational mobility have of late tended to diverge in their concerns and methodology. This reflects assumptions regarding the increasing part played by education in intergenerational mobility and the decreasing part played by class origins in intragenerational mobility, once education is controlled. The paper contributes to the questioning of these assumptions on empirical grounds. Analyses are made of the occupational mobility of men in three British birth cohorts over the course of their earlier working lives. We find that while educational qualifications have a strong effect on occupational attainment, this effect does not increase across the three cohorts; that class origins also have a significant effect, and one that does not decrease across the cohorts; and that features of work-life experience, in particular the frequency of occupational changes, likewise have a persisting effect, independently of both education and class origins. Secular changes in mobility processes are thus scarcely in evidence, but the analyses do provide strong indications of a cohort effect. Men in the 1958 birth cohort, whose first years in the labour market coincided with a period of severe recession, de-industrialisation and high unemployment, would appear to have experienced various lasting disadvantages in their subsequent occupational histories.

Keywords:

• occupational attainment
• social class
• education
• cohort effects

Acknowledgements

Our research forms part of the ESRC Gender Network, Project 1 ‘Changing Occupational Careers of Men and Women’, reference: RES-225-25-2001. We are indebted to Diana Kuh for access to the MRC NSHD data-set. For helpful comments on earlier versions, we thank Carlo Barone, Shirley Dex, Robert Erikson, Steffen Hilmert, Heather Joshi, Yaojun Li, Colin Mills, Walter Müller, Antonio Schizzerotto, participants in a seminar at Nuffield College Oxford in January 2010, and an anonymous referee.

Notes

¹In the economics literature Sicherman and Galor (1990) have developed an explicit model of occupational mobility which envisages that a significant part of the economic returns to education comes in the form of improved chances of occupational upgrading in the course of working life. This theory is criticised on empirical grounds in the work of Büchel and others, cited in the text below.

²For the 1946 and 1958 cohorts this information is of course already available but in the case of the 1946 cohort further work is necessary in order to bring the data into a suitable form for analyses of the kind we undertake here. In the case of the 1970 cohort, relevant data will be available in the near future.

³To check that the results of our analyses were not in any way artefactual on account of the greater refinement of the occupational categories of the earnings scale as compared with those of the status scale, we have re-run all analyses using a version of the former scale in which we collapse it to the 31 categories of the latter. No differences were found of a kind that would require significant modification of the commentary or conclusions of the present text. These and all subsequent results referred to in the notes are available from the authors on request.

⁴Men in the 1970 cohort were also exposed to unfavourable labour market conditions early in their working lives with the economic recession of early 1990s. However, as compared with the 1980s, unemployment rates remained at double-digit levels for a much shorter period – i.e., only from 1992 to 1994 – and turbulence in the labour market would seem to have been at a generally lower level. We take unemployment rates as an indicator of general labour market conditions rather seeking to focus on the effects of unemployment per se at the individual level. See further note 15 below.

⁵In considering possible changes in the effects of class origins across cohorts, we work simply with a binary, salariat/non-salariat distinction in order to keep down the number of parameters to be reported. However, while the contrast thus set up is a marked one, we do in this way tend to underestimate class origin effects.

⁶Here and subsequently where a difference in effect is claimed or implied as between the 1946 and 1958 cohorts, it may be assumed that this would be shown to be significant if one or other of these cohorts, rather than the 1970 cohort, were taken as the reference category.

⁷The motivation for the inclusion of variables in the matrix was evidence from previous research that they affect the probability of occupational change over the course of working life. For further details, see Bukodi and Goldthorpe (2009).

⁸We recognise that if men's work histories were to be considered over a longer period – i.e., with less right-hand censoring – some individuals treated under our model as having achieved occupational maturity by age 34 would appear, under this same model, as having reached maturity only at some later time. However, at the individual level, occupational maturity is an inherently probabilistic concept. Our concern is to use it to separate significantly differing aggregates.

⁹It is here that results for men whom we treat as not having yet reached occupational maturity differ most notably. In their case, the effects of level of first occupation are stronger but there is no positive effect of number of occupations previously held when using either the status scale or the earnings scale.

¹⁰The importance of treating education as time-variant is here underlined. Further analyses, available on request, show that in all three cohorts men who increased their level of qualifications after entering the labour market have significantly higher levels of occupational attainment than men whose qualifications remained unchanged, although it is also of interest that this effect is relatively weak for men in the 1958 cohort.

¹¹For men treated as not having reached occupational maturity the effects of qualifications are on this same pattern but are generally weaker in regard to both occupational status and earnings. With the status scale, the 1958 effect referred to in the text does not reach a statistically significant level.

¹²For men treated as not having reached occupational maturity, class origin effects, just like qualifications effects, are on the same pattern but weaker, and again when using both the status and earnings scales. However, with the status scale, an advantage of being of salariat background for men in the 1946 cohort still shows up.

¹³This advantage is however still present among the men in the 1946 cohort whom we regard as not having reached occupational maturity.

¹⁴It could be argued that data from birth cohort studies are not the fairest basis on which to evaluate arguments claiming secular trends as against those emphasising cohort specific effects: the latter effects are, if present, more likely to be revealed. While accepting this argument, we would at the same time note that in showing little evidence of secular trends of the kind we consider, our results do in fact reinforce those of earlier analyses based on repeated cross-sectional data (Goldthorpe and Mills 2004; Jackson et al. 2005; Goldthorpe and Jackson 2008) from which, however, cohort-specific effects could not readily be identified.

¹⁵Bukodi (2009) further shows, on the basis of a typology of occupational trajectories from labour market entry up to age 34, that men in the 1958 cohort are also distinctive in being more likely to have downward and less likely to have upward trajectories than men in the other two cohorts and also more likely to have ‘unstable’ trajectories, whether eventually upward or downward in their outcomes. Other studies based on the 1958 cohort have revealed adverse effects of the labour market conditions under which men in this cohort entered work (e.g., Gregg 2001; Bell and Blanchflower 2009) but these have focused specifically on ‘scarring’ by early unemployment in regard to later employment and earnings prospects. As observed at the outset, our wider interests are in the consequences for occupational attainment, and when number of months out of employment over the first 5 years of working life is included in the models we have presented, we find no significant effect on occupational level at maturity and only a very slight negative effect if we include months out of employment after the first 5 years of employment.