A test of the Becker-Tomes model of human capital transmission using microdata on four generations

We estimate the well-known Becker-Tomes (1986) model of intergenerational transmission of human capital. A Swedish data set which links individual measures on educational attainments of four generations, enables us to use great grandparents’ education as an instrumental variable. This approach was suggested already in Becker- Tomes (1986) but, because of the lack of data, never implemented. The identifying assumption, which holds within the Becker-Tomes framework, is that great grandparents’ education is unrelated to great grandchild’s education, conditional on the education of the parent and grandparent. We test the prediction that the structural parameter for grandparents’ education enters with a negative sign in an intergenerational regression model where the education of a child is linearly related to the education of the parent and the education of the grandparent. We fail to find empirical support for the model’s predictions.