Effects of Child Development Accounts on Childrenã¢â‚¬â„¢s Behavior Problems: Evidence from a Longitudinal Randomized Policy Experiment

Background and Objective: Theory and empirical examination have articulated different ways that asset building for children may positively affect aspects of child well-being, including social-emotional development and behavioral competencies. With greater assets for children’s future development, families have economic resources to invest in a better home learning environment and family interactions, and are more likely to be future-oriented and goal-directed for long-term development, potentially enhancing behavioral competencies. Our prior research has found that CDAs have positive effects on social-emotional development for children at age four years, and the effects are greater for disadvantaged subsamples. Using new data from SEED OK, this study examines the impacts of CDAs on behavior problems when children were entering adolescence (about age 14).

Methods: The dependent variable, children’s behavior problems, is measured in Wave 3 using 8 items from the Behavior Problem Index. Each item has three possible responses (1 = often true, 2 = sometime true, and 3 = not true). The independent variable is mother’s CDA treatment status (1 = treatment group and 0 = control group). We first perform confirmatory factor analysis (CFA) to test whether individual items served as a good indicator for children’s behavior problems and establish a measurement model. We further test CDA effects on the latent measure of children’s behavior problems using both the pre-COVID sample (N = 672) and the whole sample of Wave 3 (N = 1,705), controlling for baseline demographic and socioeconomic characteristics. Based on directional hypotheses, we apply one-tailed tests and set the significance level at p<.10. Additional analyses are conducted to check robustness of the results.

Results: The CFA analysis finds that four out of the eight items were stronger measures of children’s behavior problems (standardized factor loading greater than .50), and therefore the measurement model is established using these four items (sudden changes in mood or feeling; complaining that no one loves him/her; being unhappy, sad, or depressed; feeling worthless or inferior). Findings from the regression analysis indicated that, in the pre-COVID sample, CDAs have positive effects on children’s behavior problems with an effect size of 7% of a standard deviation (p <.10). CDA effects however are not statistically significant in the whole sample (effect size = .02). Robustness tests with different approaches to measure behavior problems consistently find positive CDA effects in the pre-COVID sample.

Conclusions and Implications: After the start of the Wave 3 data collection, the COVID-19 pandemic changed life routines of families and also affected data collection strategy and implementation. However, in the pre-COVID sample, which has not been disrupted by the pandemic, we observe a clear pattern that CDAs have positive impacts on children’s behavioral competencies. We conclude that a universal, progressive, and life-long CDA policy may reduce children’s behavior problems when they are adolescents. Future research should explore how CDA effects on children’s behaviors are associated with long-term education and economic outcomes.