Description:

Linh To (Boston University): The Child Penalty and an Age-Old Problem

Abstract: In this paper, we explicitly formulate the estimation of child penalties as an age-period-cohort identification problem, clarifying that the fundamental challenge is the non-identifiability of linear trends in the three sets of fixed effects. We derive the number of restrictions required for identification, showing that precisely one additional restriction suffices in contexts typically used for child penalty estimation. Using this framework, we empirically test, and reject, the common practice of omitting cohort fixed effects. We demonstrate that biases from omitting cohort effects can be substantial and can vary across data structure and empirical settings.