|Feb 2007, Ghazala Mansuri
Recent debates on “feasible globalization” have focused on the importance of opening up international labor markets to low skill guest workers from developing countries. It is argued that income gains from such a liberalization of labor markets would be large and could contribute significantly to a reduction in inequalities of wealth and opportunity, both within and across countries. Key to this is the expectation that migrant remittances will fuel private investments in both physical and human capital in origin communities.
From a policy perspective, therefore, it is important to assess the extent to which current temporary migration induces such investment. The two papers on which this note is based use data from rural Pakistan to examine this question.1 Temporary economic migration is quite substantial. One in four rural households report at least one migrant. Most migrants also maintain close ties with their origin households, returning frequently and sending substantial remittances.
Policy research working papers 3945 and 3946 focus, respectively, on the impact of temporary migration on child schooling and health, arguably the two most important contributors to inequalities of opportunity. Low educational attainment and poor early childhood growth have been viewed as arising, at least partly, from incomplete or absent credit markets and substantial uninsured income risk. Migration induced transfers should release such constraints, allowing poor migrant households to undertake larger investments in the schooling and health of their children. Furthermore, evidence suggests that the gender allocation of resources can be quite sensitive to income and that gender differences in the relative burden of risk may also be important. In contexts where son preference is significant, therefore, remittances should also reduce gender inequalities in these outcomes.
Migration can, however, also create new constraints and/or change preferences in a direction which dampens or even reverses this potential enhanced investment. Migration induced changes in family structure are of particular interest in this regard. Migration disrupts family life in any number of ways, many of which may place a greater burden on children. If, for example, mothers need to spend more time on household production, increased childcare and housework responsibilities may fall on girls. In contexts like rural Pakistan, where female seclusion practices often restrict the mobility of women, male absence may also place a greater work burden on older boys or make households more vulnerable to social exclusion or opprobrium, which may impact the type of work women in such households can undertake or the restrictions placed on young girls.
Conversely, though, 'male absence' could also change the balance of preferences over child schooling or health status in another direction. Migrant households are often female headed in the period when crucial investments in child health and schooling need to be made. A substantial body of research has identified important gender differences in preferences over the welfare of children and has shown, in particular, that investments in girls tend to increase significantly in contexts where mothers exercise greater control over the use of household resources.
In sum, migration can impact household investments in human capital through a number of competing channels, whose relevance is likely to vary substantially across contexts, making it critical to understand the specific channels through which particular effects emerge.
Assessing the Impact of Temporary Migration on Child Schooling and Health
The key empirical problem in assessing the impact of migration on household behavior is that household characteristics which influence the decision to migrate are also likely to affect other household outcomes including child schooling, labor market activity, early childhood growth and any observed gender disparities in all of these.
The papers employ two strategies to deal with this potential endogeneity problem. The first is to instrument for migration.2 This is done by using household level census data for each village to generate the village migration prevalence rate within different landownership groups. Since land is mostly inherited in this context, this yields an instrument that varies within the village but is not likely to be affected by household behavior. Further within village variation in this instrument is obtained by using the fact that mobility restrictions on women typically require the presence of an adult male in the household. Thus households with a single adult male are extremely unlikely to undertake migration. The advantage of an instrument that varies within a village is that the effect of unobserved community characteristics, such as local labor market conditions or school quality, which could affect the returns to schooling or child health, as well as the propensity to migrate, can be directly controlled for using a village fixed effects specification.3
The second strategy is to compare siblings within migrant households by exploiting the fact that many schooling decisions are time sensitive. In the context we study, children rarely begin formal schooling after age 9. We can test, therefore, whether children in migrant households who had turned 9 before the household's first migration episode are less likely to be enrolled in school as compared to children who turned 9 after the first migrant left the household. Similarly, since early childhood growth is a strong predictor of adult height-for-age, we can compare siblings born before and after the first migration episode for the household. Differences in the nutritional status of children born before and after migration, within the same household, should be free of any bias due to time invariant unobserved household characteristics.
The paper uses two measures of early child growth: weight-for-age (WAZ) and height-for-age (HAZ) z-scores.4 Kernel density estimates of the distribution of WAZ and HAZ for migrant and non-migrant households by gender indicate a positive and significant effect of migration on early childhood growth for both boys and girls (see Figure below for HAZ).5 Instrumental variables estimation confirms this and indicates significantly larger effects for girls. Estimation on samples disaggregated by age indicate that young girls have a significantly larger height advantage than do boys and this advantage is sustained as girl’s age, affirming the potential inter-generational benefits of averting nutritional and other health shocks in early childhood. These results are further validated when the sample is restricted to migrant households and the growth outcomes of siblings born before and after migration are compared.
Migration also has a significant and positive effect on all schooling decisions and a strong dampening effect on child labor market activity. This suggests that the opportunity cost of time spent by children in school may be quite substantial, at least for some households.
Most significantly, households make much larger investments in the schooling of girls, leading to a significant reduction in gender gaps in school enrollment, retention, and attainment. In contrast, there are no gender differentials in labor market activity. Thus the smaller relative gains for boys do not appear to reflect higher labor demands. A comparison of siblings further corroborates these results.
In order to understand the extent to which observed gender differentials in schooling and health could be induced by changes in household structure, we ask whether girls do relatively better in migrant households with a female head. The results are quite surprising. Female headship appears to have no additional impact on school enrollment for either boys or girls. However, adolescent girls in such households are much more likely to drop out of school and consequently have much lower school attainment. The opposite is true for boys, who do significantly better in female headed migrant households. Is this because female headship has a differential impact on the work burden of children? The data do not support this. While both boys and girls work substantially more in female headed households, there are no gender differences in work burden. There are also no gender differences in health outcomes. We also find no impact of female headship on the woman’s own labor supply. This suggests that the positive effects we observe do not derive from female headship but rather from the income effects of migration. If anything, female headship appears to protect boys at the cost of girls. The context suggests that this may be due to an increase in social vulnerability rather than preferences, since there are no negative enrollment effects but adolescent girls are much more likely to be withdrawn from school.
1 The Pakistan Rural Household Survey (PRHS) 2001-02.
2 Estimation using instrumental variables aims to identify the exogenous variation in migration. A plausible instrument should be correlated with migration but should be able to influence child schooling or health only via its impact on the decision to migrate.
3 Conditional on appropriate household demographic characteristics and village fixed effects, it is shown that the number of adult males exercises no influence on any outcome of interest.
4 Weight for age measures underweight and is the most commonly used measure of child nutritional status. It is a particularly good indicator for children under two. Height for age is a measure of stunting or chronic malnourishment. It is meant to capture the long-term, cumulative effects of inadequate nutrition and poor health status. Studies have shown that children experiencing slow height growth perform less well in school, score poorly on tests of cognitive function, and have poorer psychomotor and fine motor skills. Taller women also experience fewer complications during child birth, have children with higher birth weights and face lower risks of child and maternal mortality. Thus growth faltering in young girls may have inter-generational consequences.
5 The distributions for children from migrant households lie to the right of those for non-migrants and the Kolmogorov-Smirnov (KS) test for the equality of distributions by migration status rejects the equality of the two distributions for both girls and boys with p-value