The Causal Effect of Education on Wages Revisited*
使用早期吸烟和法定离校年龄提高作为工具变量，估计英国教育对工资的因果效应，发现早期吸烟是强且有效的工具变量，估计结果更接近平均处理效应。
Abstract This study estimates the return to education in Britain using two instrumental variable (IV) estimators: one exploits variation in schooling associated with early smoking, the other uses the raising of the school leaving age; both affect a sizeable proportion of the sample. Early smoking is found to be a strong and valid IV and unlike previous IV strategies uses variations in education at numerous points across the distributions of (i) education, and (ii) ability. Thus whilst still a ‘local average treatment effect’ the estimate is closer to the average effect of additional education, akin to least squares but corrected for endogeneity.