Identification of Child Penalties

A growing body of research estimates child penalties, the gender gap in earnings effects from parenthood, using event studies that normalize treatment effects by counterfactual earnings. I formalize the identification framework underlying this approach and show it fails to identify its target estimand when parallel trends in levels are violated. Insights from human capital theory suggest such violations are likely: higher-ability individuals tend to delay childbirth and experience steeper earnings growth. This selection mechanism suggests conventional estimates understate child penalties for early-treated parents. I propose instead to target the effect of parenthood on the gender earnings ratio and show it is identified without further assumptions. Based on Israeli administrative data, estimates of this alternative are 30% smaller in absolute value than conventional estimates for later-childbearing parents. Bias-corrected estimates, bounding fathers’ effects within a reasonable range, suggest conventional estimates understate child penalties by 25-50% for early parents.