*Names in bold indicate Presenter
This is not the first paper to explore the implications of AYCE policies for maternal employment and child well-being. Recent examples include Hill (2012) and Washbrook, Ruhm, Waldfogel, and Han (2011), both of which measure AYCE policies through state-level dummy variables indicating the number of exemption-months allotted. However, this coding masks important individual-level variation in how these policies are implemented. For example, some states (e.g., Georgia and Maryland) provide a comparatively long exemption period for women’s first child and a significantly shorter period for each subsequent child. Other states (e.g., Washington) provide a lifetime allotment of exempt time that can be used for more than one child.
This project carefully codes these complex AYCE policies, exploiting both cross-state and birth-order variation in exposure to different amounts of exempt time. This information is then merged to children in the Birth cohort of the Early Childhood Longitudinal Study (ECLS-B), a nationally-representative cohort-based study of about 10,000 children born in 2001. The ECLS-B contains high-quality information on the focal child’s state of residence, birth-order, and age-at-assessment, which are used to construct a measure of the potential number of months remaining in the mother’s exemption from the welfare work requirements. Negative values indicate that a mother has not yet hit the AYCE limit, while positive values indicate that she is over the limit (median: 5.4 months; min: -16.8; max: 21.8).
Using a sample of children of disadvantaged mothers (9-month survey; N=6408), I estimate first-stage regressions of early maternal employment (any work and months of work) on the AYCE variable, controlling for state characteristics, child birth-order, and age. I find that longer AYCE time allotments are strongly negatively correlated with maternal employment (F-stat≈20). Reduced form regressions indicate that longer time allotments are associated with higher cognitive ability test scores for children at 9- and 24-months. Given these results, it is not surprising that the instrumental variables models, which use AYCE time allotments as an instrument for maternal employment, indicate that first-year maternal employment reduces test scores at 9- and 24-months. An additional three-months of work lowers test scores by 0.23 SDs (preliminary).
Findings from this research highlight an unintended consequence of welfare work requirements (and thus have implications for the optimal design of AYCE policies), and may also contribute to the policy debate on the potential importance of job-protected parental leave systems.