Review of Previous Studies

Browning (1992) reviews the US literature on the impact of children on female labor supply up to the early 1990s. The observation he made was that it is clear that there is a strong negative correlation between the presence of young children and the labor supply of their mothers measured either as labor force participation or labor hours. The relationship, however, is not as clear if one moves from correlation to causal relationships. This is because the relationship between work and child bearing is very complex conceptually and also to estimate empirically.

One important point highlighted in Browning (1992) is that labor supply equations that do not include the child variables as regressors are by implication estimating a reduced form. To illustrate the importance of recognizing the endogeneity of fertility the review indicated that those that consider fertility as endogenous have yielded not significant or even positive relationships putting into question results that show a negative relationship. He expressed frustration that the studies that considered the endogeneity of fertility did not also show OLS results to be able to make a comparison. The results in Rosenzweig and Wolpin (1980) show that not instrumenting for the fertility variables underestimates the negative impact of exogenous changes in fertility. The results in Angrist and Evans (1998), on the other hand, show that controlling for endogeneity either by the use of the sex of the first two births or twins as instruments yields a negative impact of the number of children on the labor supply of mothers, but the impact is much more subdued than that obtained from OLS estimates or the opposite of what Rosenzweig and Wolpin (1980) obtained. Gangadharan and Rosenbloom (1996) add that the negative impact on hours increases with the magnitude of mothers entering the labor market and that while the negative impact on earnings in 1980 is temporary, the 1990 results show continued depressed earnings even after the labor supply effects had disappeared. Thus, while the negative effect of children in mother’s labor supply is found in these three studies, the direction of the impact of recognizing endogeneity of fertility variables in the labor supply equation is not clearly established. In the case of the father’s labor supply, Angrist and Evans (1998) did not find a significant effect of children. Lundberg and Rose (2002), however, found a positive effect of children. In addition, they found that the effect of children on male labor supply is substantially larger when endogeneity is taken into account lending support to the Rosenzweig and Wolpin (1980) result. Vere (2005) adds the new 2000 Census data to the 1980 and 1990 data used in Angrist and Evans (1998). He finds that men respond to additional children by increasing earnings, lending support to Lundberg and Rose (2002). It is also noted that specialization in husbands’ and wives’ roles in raising children has become more pronounced over time. In the case of the husband, not only is the sign of the impact of children not clear, but also the direction of the effect of controlling for the endogeneity of the number of children is unclear.

If the relationships between children and the parents labor supply and earnings are not clear in developed county literature, the same is true for developing countries with the little evidence that is available.

The only study, to the knowledge of the author, using Philippine data that considers the endogeneity of children in women’s work hours and earnings is Adair, et al. (2002). Using the Cebu Longitudinal Health and Nutrition Survey from 1983-1991, the paper models the joint decision of sector of work choice (wage, piece or self-employed) and earnings (or hours of work) with children included in the latter equation. The study finds that children negatively affect hours and earnings in what the authors deemed as a form of a “child tax”. They find that for the change in earnings equation the number of additional live births has a significant effect but the number of children less than 2 years is not. For the change in labor hours equation, it is the number of children less than 2 years that is significant. Child bearing, however, was not found to affect sector of employment.

The review of other studies using Philippine data that assume fertility variables as exogenous show mixed results. Using survey data from Laguna province, Quizon-King (1978) found that while the number of children does not significantly affect the market time of both mother and father, the presence of children 6 years and below increases the home production time of mothers. Garcia (1990) also found that the percentage of children below six years old significantly increases the time for home tasks and decreases market work for wives, while the opposite is true for husbands. Popkin (1983) found that the number of children below 6 yeas old decreases the leisure time of rural mothers; the presence of children one year old and below decreases their market production time, while the presence of children 1-6 years old increases their home production time. King and Evenson (1983) found that the presence of children six years old and below increases the home time of wives. For husbands, the presence of children one year and below positively affects home time. When the age structure of the children was controlled for the effects turned out to be small. Market time, however, of both husband and wife, is not affected by the presence of children. Finally, using cross tabulation analysis from the Cebu data Tiefenthaler (1997) has confirmed earlier results on the negative relationship between child bearing and labor supply of mothers. However, it was pointed out that that for mothers with previous children, labor market hours 14 months postnatal is the same as prior to the sample birth, although this is lower for first-time mothers. For fathers, birth does not significantly affect their labor market hours. In addition, even if mothers decrease labor supply within the year after birth, fathers do not increase their market time to compensate for lost income. Birth only slightly increases father’s time in childcare and only if there are no other children. It was also found that births increase the childcare and market time of older daughters (13-17 years). Thus, it appears that older daughters, rather than fathers, substitute for the lost market time of mothers due to birth.

Evidence from other developing countries also has varied results. Using the 1992 DHS for Morocco Assaad and Zoari (2003) found that in urban Morocco the presence of school-age children significantly reduces participation of women in all types of wage work. This particular study uses a three-stage sequential modeling of marriage, child bearing and labor force participation with predicted values of the dependent variable used in the subsequent stage to control for endogeneity. Multivariate analysis using data in urban Pakistan and assuming fertility variables as exogenous showed differential effects by sex of children (Cochrane, Kozel and Alderman, 1990). Males 7-14 and over 14 reduce the participation of women while females 7-14 and over 14 significantly increase the labor force participation of women. The magnitude of the effects, however, was small. The authors added that these implies that males provide an alternative source, while females free up women’s home time. In the case of rural Thailand, Poshisita et al (1990), using cross tabulation analyses supported by results from focus group discussion, find that while children do not prevent rural Thai women from working, they interrupt work and interfere with economic activity.

Download this Discussion Paper [ PDF 268.9KB| 27 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.