Search

Using the 1991 to 2007 waves of the UK British Household Panel Survey (BHPS), the authors estimate a fixed effects specification that has as outcomes (i) earnings and (ii) an indicator of social position measured using the CAMSIS scale. Adopting a fixed effects specification enables them to isolate the role of lifelong learning on these two outcome measures, as distinct from other person-specific variables which might be correlated with wages and education level. This study of the BHPS also adds an important dimension to the growing literature within the UK, where evaluation of the returns to lifelong learning has been very much focused on vocationally-oriented programmes (including National Vocational Qualifications or NVQs) and based on analysis of cohorts (Dearden et al., 2002; Jenkins et al, 2003; Dearden et al. 2004; De-Coulon and Vignoles, 2008). The evidence arising from these UK studies of the returns to lifelong learning is mixed. De-Coulon and Vignoles (2008) find the strongest evidence that low level vocational qualifications (NVQ2) increase earnings for women. Jenkins et al (2003) identify a positive impact of lifelong learning on employment probabilities, with earnings impacts limited to men who left school with low level qualifications and subsequently obtained a degree. However, there is also evidence that some individuals who gain low level vocational qualifications (NVQs at levels 1 and 2) have lower earnings than those without these qualifications (Dearden et al. 2002). Dearden et al. (2004) and Greenwood et al. (2007) detail how the estimated returns to these NVQ qualifications vary for sub-groups of the population and across sectors/occupations. Whilst Dearden et al. (2002) utilise the panel element of the Labour Force Survey in addition to analysis of cohort studies, this only allows consideration of individuals over a one year period and the authors are unable to account for unobserved heterogeneity. The use of the BHPS allows the authors to consider (i) long-term returns to lifelong learning (ii) that do not suffer from the potential for simultaneity that arise when one estimates returns using cohort studies. Their use of a distributed lag structure allows clear identification of whether human capital accumulation takes place before changes in earnings or other outcomes measures, in comparison to cohort studies. However, these substantial benefits do represent something of a trade-off, as analysis of the BHPS also has some limitations. Primarily, these relate to the limited level of disaggregation by qualification type and personal characteristics that is possible; arising from both a problem of sample size and also issues with the categorisation of educational attainment within the BHPS. However, in this respect this paper serves to highlight the data-needs that surround the development of new panel data sets, for instance the proposed Longitudinal Household Survey within the UK. (Contains 2 figures, 9 tables and 2 footnotes.)