Do Maternal Investments in Human Capital Affect Children’s Academic Achievement?⊗ Quinn Moore Department of Economics University of Michigan Lucie Schmidt Department of Economics Williams College May 2004 ⊗ Katie Vitello provided excellent research assistance. We are grateful to Richard Arum, Paula Baker, Rebecca Blank, Katalin Bognar, Charles Brown, Andreas Pape, Viji Sathy, Purvi Sevak, and Heidi Shierholz for useful comments and suggestions. All errors and omissions are our own. Moore wishes to acknowledge financial support from the National Science Foundation.
Corresponding Author: Lucie Schmidt, Department of Economics, Fernald House, Williams College, Williamstown, MA 01267; lschmidt@williams. Do Maternal Investments in Human Capital Affect Children’s Academic Achievement? Abstract Children of educated mothers fare better on a variety of educational outcomes. However, little research has been done on the effects of human capital investments undertaken by mothers with children at home. Such investments have a theoretically ambiguous effect on child outcomes, since human capital investment reduces time spent with children but may have positive spillover effects on child investment.
Using child- and sibling-fixed effects models to deal with unobserved heterogeneity, we find that cumulative maternal schooling undertaken during a child's lifetime has significant positive effects on child outcomes, and that negative time allocation effects are minimal. Introduction Maternal education level is consistently found to be positively correlated with children’s cognitive development and educational outcomes. In 1998, the National Center for Education Statistics (NCES) reported that of first-time kindergartners, 16% of children with high school graduate mothers scored in the highest quartile on reading. For college-educated mothers, this percentage was 46%.1 These strong correlations seem to suggest that from a policy standpoint, encouraging mothers to acquire additional education may positively affect children’s educational outcomes.
However, once time inputs are considered, the theoretical predictions of the effect of a mother’s return to schooling are ambiguous. In this paper, we seek to answer an important policy relevant- question: What is the relationship between the educational investments of mothers and their children’s outcomes? There are several mechanisms through which parental educational decisions might affect child outcomes. The human capital investment model (Becker, 1981; Becker & Tomes, 1986) focuses on the key insight that resources within the family are limited, and that parents make decisions regarding the level of resources to invest in their children versus the amount to spend on consumption. This approach suggests that highly educated parents, with greater levels of income, may invest more in their children’s education.
In addition, highly skilled parents may be more productive in parenting, 1 One possible mechanism for this effect is genetics, but differences in inputs to the parenting process may also be important. For example, NCES also reported that 39% of kindergartners with high school graduate mothers were read to every day, as compared with 59% of kindergartners whose mothers had a Bachelor’s degree or higher. 2 generating larger increases in child human capital for a given cost. Alternative theories suggest possible role model effects, where the transmission of parental behaviors to children is through socialization (Woelfel and Haller, 1971).
As would be predicted from these theories, the existing empirical evidence consistently finds that higher levels of completed parental education are strongly and positively associated with child outcomes. For young children, higher levels of completed parental education are significantly positively correlated with higher standardized test scores (Liebowitz, 1977, Blau and Grossberg, 1992). For older children, high school graduation and years of schooling completed are positively and significantly affected by the level of completed parental education.2 Furthermore, maternal education is generally found to have a larger effect than father's education on children's educational outcomes (Haveman and Wolfe, 1995). A conclusion frequently drawn from these and related findings on the relationship between maternal education level and child academic achievement is that children would be well served if mothers were encouraged to obtain additional education.
In a recent chapter of an edited volume on investments in children, Lisa Lynch argues: [O]ne of the most important determinants of education is parental education. If we want to raise education levels in the United States, we need to consider investments in both youths and their parents, recognizing that parents are teachers too. Raising the skills and education of incumbent workers not only makes them more productive in the workplace but also contributes to the education of their children (Lynch (2000), 43-44). 2 See Haveman and Wolfe (1995) for a survey of this literature.
3 Implicit in this policy prescription is the assumption that human capital investments undertaken by women while they are mothering children will have positive effects on family outcomes. It is not clear that this assumption is justified by theory. A simple model of household time allocation would suggest that time spent on human capital investment would reduce time spent with children, and thus negatively impact child outcomes. Alternatively, maternal time in education and training activities may have positive spillover effects on investment in children, including role model effects, improved ability to help with homework, and improved ability to navigate the educational system.
In addition, increased family income associated with the attainment of higher levels of maternal education may also have a positive effect on children’s outcomes. It is worth noting that the negative time allocation effects associated with maternal human capital investment are likely to be strongest in the contemporaneous period, while the positive spillover effects are likely to be less transient. These competing theoretical effects provide ambiguous predictions for the relationship between maternal human capital investments and child outcomes. Similar ambiguous predictions exist regarding the effects of maternal labor force activity.
Children may suffer as a result of increased time spent away from their mothers. However, increases in income, as well as positive role model effects associated with maternal labor force involvement, may be beneficial for children. While a large and growing literature has focused on the effects of maternal employment on child outcomes, there has been little attention paid to any similar effects of maternal schooling. 4 In this paper, we use data from the Child Supplement of the National Longitudinal Survey of Youth 1979 (NLSY79) to examine the effect of a mother undertaking schooling on her children’s academic outcomes.
In addition, we compare the effects of maternal schooling to effects of maternal labor force participation. Using both child- fixed effects and sibling-fixed effects models to deal with unobserved heterogeneity, we find evidence that negative effects due to decreased time allocated to children appear to be minimal. We also find that cumulative maternal schooling has significant positive effects on child outcomes. The remainder of the paper proceeds as follows: Section II discusses the theoretical and observed empirical relationships between maternal education, maternal employment, and children’s outcomes.
Section III explains our methodology, and Section IV describes our data. Section V presents results, and Section VI concludes. Children's Outcomes and Maternal Education and Employment Despite the strong positive relationship between child academic achievement and maternal educational level, the relationship between academic achievement and maternal educational investment is ambiguous. We develop a theoretical model that shows that under certain conditions, mothers who value their own consumption and the final level of their children’s development will choose to go to school during their child’s lifetime.
We then discuss the theoretical implications of the maternal schooling decision for the child’s development level. Theoretical Model 5 Our theoretical framework is an adaptation of the framework developed by Stafford (1987) in modeling the tradeoff between women’s market career and home career when they value the development of their children. In particular, we focus on maternal skill correlated with education and add this “maternal knowledge level” to the child’s development function. Children’s development, K , follows K = Q ( ct , M t , t ), Qc > 0, Qcc < 0, (1) QM > 0, QMM < 0, QcM > 0, Qt < 0, where Q (⋅) is the function relating child development to maternal time spent on child, ct , and child’s age (time since birth), t.
The portion of skill that is correlated with educational attainment, M t , enters positively into the child development function both directly ( QM > 0 ) and through an increase in the efficiency of time spent with children ( QcM > 0 ). Maternal skill from education, or maternal knowledge level, changes according to M t = f ( st ) − δM t , (2) f s > 0, f ss < 0, where st represents maternal time invested in education, f (⋅) is a function relating this investment to knowledge levels, and δ is a depreciation rate. Total time is divided among market work, school, and childcare: (3) 1 ≥ lt + st + c t. 6 Parents’ welfare is a function of earnings and the child’s development state at the end of the planning period; T V = ∫ e − ptαlt M t dt + K T , β (4) 0 where T is the end of the planning period, p is the discount rate, and α is the per unit wage rate.
Thus αlt M tβ represents the income earned through labor market participation.3 Child’s development level at the end of the planning period, KT , is the result of the accumulation of development given by equation (1), T (5) K T = ∫ Q ( ct , M t , t )dt. t t This maximization leads to first order conditions (7) λ M f s = Qc = e − ptαM β where λM is the shadow value of knowledge at time t, which follows (8) λM = − e − pt βαlt M tβ −1 − Q M + δλM. We explore the implications of this solution by restricting the functional form of Q (⋅) and f (⋅). In particular, we let child development evolve according to 3 Here we make the simplification that production only depends on skill related to education.
A more complicated skill set adds unnecessary complexity to the model. This equation gives us decreasing returns to both maternal time spent with children and maternal knowledge level. This is a simplification in that the returns to the inputs are not dependent on the child’s age, nor do the returns to maternal time spent with children increase with maternal knowledge level (i. the cross derivative is not positive).
We let maternal time spent on schooling increase maternal knowledge level according to (10) f ( st ) = ϕ ln( st ), 0 < ϕ < 1. This equation has the desired property that knowledge level increases at a decreasing rate with maternal time spent on education at time t. Using the first order conditions, these functional restrictions give us the following time paths for maternal time spent on child care and education: γ pt − β (11) ct = e Mt , α ϕ pt − β (12) st = e M t λM. α The state and costate variables then follow ϕ (13) M t = ϕ ln( e pt ) − ϕβ ln( M t ) + ϕ ln(λ M ) − δM t , and α (14) λM = − M −1 ( e − pt βαM tβ − βϕλ M − βγ + σ ) + δλM.
Which leads to isoclines of δ α β ϕM λ M = e − pt t (15) M e , and ϕ t 8 e − pt βαM β + σ − βγ (16) λM =. δM t + βϕ We derive the intercepts and the derivatives of these isoclines in the appendix. Figure 1 σ presents the phase diagram in the case where > γ .4 An individual with initial β knowledge level M 0 , which is lower than the level of knowledge where the isoclines intersect, will invest in additional education for a period before allowing her knowledge to depreciate.