Paid Parental Leave and Maternal Health: A Difference-in-Differences Study Using PRAMS 2012–2022

• In Event: Support or Strain? The Real-World Impact of Family Policy on Health, Care, and Work

Abstract

Background: This study uses Pregnancy Risk Assessment and Monitoring System (PRAMS) data (2012-2021) alongside state-level paid family leave policy enactment data to examine the effect of paid parental leave (PPL) policy on key maternal health outcomes among low-income individuals. Lower-income mothers may benefit more from a mandatory state-level PPL policy given high wage replacement rates promoting uptake.

Methods: We use difference-in-differences (DID) to estimate causal effects of PPL enactment on breastfeeding duration, postpartum checkup, and postpartum depression in New Jersey and Washington State. Both states enacted 12 weeks of PPL at up to 90% wage replacement in 2020. Data from 17 states were used to construct a counterfactual. We used multiple imputation to estimate single (non-consecutive) missing years of data among 5 states. The final analytic sample includes 19 jurisdictions across 10 years, yielding 190 location-year observations. Treated states include New Jersey and Washington State and control states include 17 states without PPL during the study period.

We estimate a two-way fixed effects (TWFE) model, controlling for state and year without concern for heterogeneous treatment effects as PPL in the states examined occurred simultaneously. Valid DID estimates require parallel trends in the absence of treatment along with no anticipation effects. We did not observe violation of parallel trends in the pre-treatment period. Falsification checks will evaluate violations of non-anticipation.

Because the TWFE approach is limited in its ability to control for time-varying confounding, we employ a trajectory balancing approach, using a kernel mean-balancing technique that assigns weights based on pre-intervention characteristics to create a synthetic counterfactual of the treatment group based on comparison states. This approach accounts for time-varying confounding and has the added benefit of relaxing the linearity in prior outcomes assumption.

Results: Results based on the TWFE models, adjusted for pre-treatment covariates, show that, among low-income women who recently gave birth, PPL enactment increased postpartum visit attendance (ATT 0.0433 [CI 0.0016, 0.0849]). Although not statistically significant, our analysis suggests that PPL enactment may have increased breastfeeding duration slightly (ATT 0.0141 [CI -0.0154, 0.0437]) and also may have led to a slight reduction in rates of postpartum depression (ATT -0.0237 [CI -0.0494, 0.002]). For trajectory balancing, we include years 2017-2021, as the approach is optimized with fewer pre-treatment years. Preliminary results based on this method suggest trends similar to the TWFE model, pointing to a slight increase in the probability of a postpartum checkup, increased rates of breastfeeding duration, and a slight reduction in incidence of postpartum depression. Further analysis will use both mean and kernel balancing approaches, along with comparison state weights, to conduct lagged regression to obtain standard errors for all estimates.

Conclusions: PPL policies may increase postpartum visit uptake among low-income mothers. Small positive effects on breastfeeding duration and a small reduction in postpartum depression were observed, though were not statistically significant. Future research should examine these effects among low-income mothers across additional states. 

Authors

• Amelia Ruth Reese, University of California - Los Angeles
  Presenting Author

• Keri Lintz, University of California - Los Angeles
  Non-Presenting Co-Author

• Dana Beck, University of California - Los Angeles
  Non-Presenting Co-Author