Jan Marcus and Daniel Kemptner, both DIW Berlin Graduate Center
This is the first study investigating the causal effect of maternal education on child's health behavior and schooling outcomes in Germany. Our paper adds to the existing literature on intergenerational transfers in various ways. We apply a new instrumental variables approach which has not yet been used in the intergenerational context. Therefore, we draw on a rich German panel data set (SOEP) containing information about three generations. This allows instrumenting maternal education by the number of her siblings while conditioning on a comprehensive set of variables describing the grandparents' social status. Given the grandparents' social status, the number of siblings generates exogenous variation in the years of education by affecting household's resources which are available for human capital investments per child. Unlike studies investigating returns to education by exploiting policy changes, our methodology works also for cohorts unaffected by policy changes and for the limited sample sizes of common household panels. We estimate the causal effect on a wide range of child outcomes both at birth and at age 18/19. We present evidence for strong and significant effects on schooling outcomes for both sexes. And, we find substantial effects on health behavior for daughters at age 18/19, but not for sons. Our findings are robust over a variety of different specifications. We also discuss assortative mating and household income as possible channels of causality. Though, we do not find evidence that the effect of maternal education works through these channels.