Govert Bijwaard , Netherlands Interdisciplinary Demographic Institute (NIDI)
Kieron Barclay, Stockholm University
Despite a substantial association between education and mortality the causal interpretation of this relation has been challenged. This association may be confounded by factors that influence both education and mortality. Moreover, surprisingly little research has investigated the underlying causal mechanism of education on mortality in the presence of one or more intermediate variables, such as income. In this paper we estimate the impact of education on mortality and how it is mediated by income using an inverse propensity weighting method that accounts for direct and indirect effects of education on the mortality rate (running through income development). We use data from the Swedish Military Conscription Data, linked to information on the parental socioeconomic situation at birth, the parental education, the education of the individual himself, date of death (up till 2012) and annual income for the period 1968 till 2012. We estimate four separate models for the educational gain in mortality using the data of individuals in two adjacent educational levels. For each pair we derive how much of the educational gain in the mortality rate can be attributed to the effect of education on the income development and how much can be viewed as a direct effect of education. Our empirical analyses reveal that for the low educated the educational gain (a 20% reduction in mortality) is mainly a direct effect of education, while for the medium educated the reduction in mortality for obtaining higher education runs mainly through the impact of education on income development.
Presented in Session 18. Mortality Determinants