The Dynamics of Child Care Demand

This paper analyzes changes in child care arrangements of a sample of children from the Youth Cohort of the National Longitudinal Surveys over the first three years of life. The analysis indicates that turnover in child care arrangements is surprisingly low among this sample and is more common among families of higher socioeconomic status. Child care turnover is positively correlated with mothers' employment turnover, but is not correlated with changes in mothers' marital status or additional births in the families. A Poisson regression model is estimated in which both observed and unobserved variables are allowed to influence child care turnover over time. Because there are multiple observations for the same child and for different children within a family, it is possible to distinguish unobserved mother effects, child effects, and purely transitory effects. The estimates reveal that child care turnover is highly correlated over time, owing mainly to the effects of observed variables and unobserved characteristics of the mother that persist over time.