Mothers’ time spent in care of their children and market work: a simultaneous model with attitudes as instruments

Abstract

Using a model that addresses the potential endogeneity of employment hours on mothers’ child care time and vice versa, by including instruments based on parental attitudes, we find a significant negative (but inelastic) relation between the two time uses.

Notes

¹ Discretionary time is defined as the hours remaining after biologically necessary bodily maintenance such as sleeping and eating.

² Some studies have considered the demand for child care as it depends on employment status without allowing for endogeneity (e.g.Henriques and Vaillancourt, 1988). Cheng (1996) uses a causality test with time series data and argues that the presence of small children discourages employment, but employment does not discourage fertility. Other studies have shown that child care costs affect employment hours (Connelly and Kimmel, 2003; Powell, 1998) but have not looked directly at the hours trade-off.

³ The sampling technique was similar to that reported in Dalenberg et al. (2004).

⁴ The following definition of child care time provided the basis for this question: ‘This includes only the time spent actively engaged in child care when that was the primary activity. Merely being at home with a child is NOT time spent at child care. For instance, if you are primarily doing something else (e.g. meal preparation or sleeping) that is NOT child care time.’ The definition of child care itself was important for its study. On the basis of discussions with a number of child care experts in the Missoula area, we defined child care as follows: ‘Child care is all activities which constitute the care or nurturing of a child, i.e. teaching, changing, exercising, cleaning, feeding (but not meal preparation), entertaining, transporting, nursing, coaching, disciplining, playing with, holding, monitoring, or any other activity with the purpose of enhancing the child's life or fulfilling the child's needs.’

⁵ See Bound et al. (1995) for a discussion of problems caused by weak instruments.

⁶ Hausman tests on each equation reject the null hypothesis of exogeneity leading to the conclusion that two-stage least squares is necessary for this estimation.

⁷ Using the amount of care provided by the mother instead of mothers’ hours of child care as described in Dalenberg et al. (2004) yielded similar results to those reported here.