We investigate the causal impact of education on life-expectancy using data for England and Walesfrom the Health and Lifestyles Survey and how that impact is mediated through changes in healthbehaviour (smoking, exercise, having breakfast). For identification of the educational gain in mor-tality we employ a Regression Discontinuity Design implied by an increasein the minimum schoolleaving age in 1947 (from 14 to 15) together with a principal stratification method for the mor-tality hazard rate. This method allows us to derive the direct and indirect (through one or moremediators) effect of education on the implied life-expectancy.Basic maximum likelihood estimation of a standard Gompertz hazard model for the mortalityrate suggests that staying in school beyond age 15 years significantly increases life-expectancy bymore than 14 years, with large indirect effects running through smoking and exercise. In contrast,estimates from the principal strata method indicate that the educational gain is much smaller (andstatistically insignificant) for those who were induced to remain in school beyond age 15. The directeffect of education is even negative for females (but statistically insignificant). Neither, do we findstatistically significant indirect effects of education on mortality running through health behaviour.
|Publisher||University of York|
|Number of pages||33|
|Publication status||Published - Feb 2019|
|Name||HEDG working paper|