The Causal Effect of Fertility Timing on Educational Attainment: An Identification Test Using the Longitudinal Structure of Schooling

Kevin Stange, University of California, Berkeley

The strong correlation between educational attainment and fertility timing is well documented. For instance, young mothers seldom go on to earn a college degree. Researchers often use controls for observable characteristics or sibling fixed effects to determine whether this correlation reflects causation or simply unobserved factors. Using the cumulative nature of educational investment, this paper demonstrates that the key identifying assumption behind these approaches does not hold. Mothers and observably similar non-mothers begin college on the same footing, but their paths diverge well before the former enter pregnancy. No such divergence should exist if the causal effect of childbirth is the only reason educational attainment is lower for mothers than similar non-mothers. This finding suggests that time-varying factors that cause women to leave school and then enter parenthood are also behind the correlation. Controls for pre-determined characteristics and fixed effects do not address this temporal source of omitted variable bias.