The wage returns to education over the life-cycle: Heterogeneity and the role of experience*

This paper considers the implications of using regression discontinuity methods (RDD) on pooled data to estimate the causal effect of education on earnings, in particular when exploiting changes in minimum school leaving age requirements. RDD methods estimate the return to schooling inclusive of the impact of lost (potential) labour market experience as a result of being in school for an additional year (i.e. the “net” return). This is in contrast to the more parametric approaches to instrumental variables estimation used in the literature of the 1990s and 2000s, which estimated the “gross” return, conditioning out differences in (potential) experience. This has implications for interpreting the returns to schooling and we demonstrate the importance of the distinction using the 1972 Raising of the School Leaving Age (RoSLA) in England and Wales. Using data from the New Earnings Survey Panel Dataset 1975-2011, we illustrate the effect of the lost labour market experience at the RoSLA discontinuity on estimates of the return to education. When the “experience penalty” is not taken into account, RDD estimates suggest a significant negative return to additional education whilst when the experience component is taken into account, a positive effect is found, in line with earlier parametric studies. Moreover, we show that there is substantial variation in the impact of RoSLA over the lifecycle, which is often masked by studies that estimate the return over many years of pooled data. Together these factors explain the range of estimates of the return to education in the UK literature.