Search Results

Method: We utilize the National Longitudinal Survey of Youth 1997 (NLSY97), which is the only national dataset that provides detailed reliable information on family wealth, SAT score, and college attendance. The study sample drawn from the NLSY97 consists of children who reported that they had taken the SAT between 1996 and 2003 (N=3,216). We construct wealth variables measuring illiquid assets, liquid assets, and debt. For statistical analysis, we use structural equation modeling (SEM) with SAT scores (verbal and math) serving as indicators of scholastic ability. SEM enables us to simultaneously estimate both direct and indirect effects on college attendance. Results: The SEM results support the ability hypothesis rather than the borrowing constraints hypothesis. Although, as expected, the liquid and illiquid assets have a positive effect, the direct effects are not significant. On the other hand, a family wealth measure, liquid assets shows a meaningful, indirect effect on college attendance through SAT (Odds ratio: 1.014, p<.05). Finally, although it is not of primary interest, family income, a traditional way to measure economic resources, is found to have a direct effect on college attendance (Odds ratio: 1.074, p<.05).

Conclusions and implications: This study suggests that family wealth, especially liquid assets which are easily convertible, is important for children's cognitive development, which is a strong predictor of post-secondary education. This finding suggests that it is important to enhance programs designed to encourage poor families to accumulate assets to promote educational achievement of their children. It should also be noted that given the observed direct effect of family income on college attendance, this study does not contradict current policy which helps youth from lower income families finance post-secondary education through borrowing.

Method: Using data from the National Longitudinal Survey of Youth 1979 (NLSY79) and its Child Supplement (NLSY-CS), we pooled 7838 children born to the NLSY female sample. Parental shift work was measured by five categories: 1) day shifts (if the main job begins at 6 am or later and ends by 6 pm); 2) evening shifts (between 2 pm and midnight); 3) night shifts (between 9 pm and 6 am); 4) other shifts (i.e., split-shift, rotating shift, and irregular hours); and 5) not working. Children’s cognitive outcome was measured by the Peabody Picture Vocabulary Test - Revised (PPVT-R). We conducted OLS regression followed by the SFE that regressed differences in sibling PPVT-R scores on differences in siblings’ exposure to parental shift work, differencing out any sibling-invariant characteristics associated with the family, including any unobserved heterogeneity that is constant across siblings within the family.

Results: Our OLS model suggested that paternal night shifts had a negative effect on children’s PPVT-R, while no maternal shift works had significant impacts. Specifically, the PPVT-R score of children with fathers working night shifts was on average about 7 points lower than that of their peers with fathers working standard day shifts (b=-7.36, p<.01), indicating that children with fathers working night shifts fell .35 standard deviation on the PPVT-R scale behind. However, in our SFE model, this negative effect of paternal night shifts dramatically decreased and was not statistically significant (b=-1.88, p>.05), indicating that there was essentially no difference in the PPVT-R between paternal night shifts and standard day shifts.

Implications: Unlike previous studies, our study does not provide empirical evidence that parental shift works have negative effects on children’s cognitive outcomes, suggesting that we should be cautious in making a causal claim from observational data. However, given that the SFE also has limitations, further research is needed to ascertain the causal nature of the intergenerational effects of parental shift works.