Why would upward trends in schooling make a nation healthier? The case of smoking in Twentieth Century France

as absolute levels in explaining the education-health gradient. We show that relative education impacts smoking, when direct utility is relative, or when there is signalling in the labour market. We use data from the "Enquête sur les Conditions de Vie des Ménages 2001" and a major reform of the education system, the Haby reform, to test the competing hypotheses. Descriptive statistics show that education has more effect on the decisions to start and quit for the birth cohorts affected by the reform. However, duration analysis reveals that, controlling for changes in policies, this result holds only for quitting.