Colección de grupos difusos de documentos o sentencias

Our results outlined in Model 3 (see Appendix, Table A3.1.) indicate that scheduled tribe children are less likely to enrol in school in UP. In UP, both scheduled caste and other backward caste children are less likely to enter into school, while it is scheduled tribes and other backward castes that are less likely to go to school in AP. Muslim children, similarly, are less likely to go to school in both AP and UP. We have already

seen that mother’s education influences female autonomy and through this affects SSA. Our results in Appendix 2.3. indicate that mother’s education also has a direct significant positive effect on the hazard of the child entering school in AP, though this effect is variable across models in UP. Father’s education is also highly significant and this is true for both primary and secondary education, though the size of the latter effect is smaller.

Wealth has a significant effect in UP, particularly in the top quintile and especially so in AP in all quintiles. It is clear that the probability of school entry is higher for the 5th wealth quintile than for the first in all the models for both AP and UP. The conditional probability of a child entering school is higher in rural areas in UP and AP than in urban areas, which is rather surprising given that schools are further apart in rural areas and there are more alternative uses of children’s time in farming regions.

The birth order of the indexed child slightly increases the probability of school entry for the second born child in AP. However, in UP, children born second or third show lower probabilities of entering into school than firstborn children. Contrary to this, the influence of the birth order being three or above in UP on the hazard is positive and significant in Model 3. Furthermore, girls appear to be significantly less likely to enrol in school in UP and significantly more likely to enrol in AP. With regard to the initial conditions, we find these to be highly significant in UP. Children who were 6 years in 2003 were more likely to go to school in UP whereas those who were 6 years in 2005 were less likely to go to school. In AP, children who were 6 years in 2004 and 2005 were less likely to go to school.

We can conclude that both in AP and UP, socio-economic characteristics of the household and demographic characteristics of the child as well as initial conditions play a significant role in determining the age of entry of the child into school.

2.8. Discussion and Conclusion

According to conventional wisdom, women’s autonomy is higher in the Southern compared to the Northern states. UP, for instance, is known for its particularly skewed sex ratio. With a national average of 106 men for 100 women, UP has 112 men per 100 women. AP shows a ratio below the national average of 102 men per 100 women.50 AP lies within the Southern kinship system but is considered the median Indian state on a number of indicators. Contrary to this, the indices constructed by adding the qualitative answers provided by the woman within the NFHS (Table A2.5.) suggest that Northern states are characterised by higher female autonomy than the South. UP shows an overall female autonomy index of 14.8. By contrast, the index for AP is 12.8. Our more detailed analysis offers greater insight into these findings. In particular, two results stand out clearly distinguishing the experience of female autonomy as well as school entry in AP and UP. In AP and UP, almost all socio-economic characteristics (caste, religion, birth cohort, rural abode) influence female autonomy. This confirms that in these states female autonomy depends heavily upon the socio-economic characteristics of the woman. It is also possible that additive indices customarily used, fail to properly reflect the autonomy enjoyed by women, in particular because they neglect the multi- dimensionality of female autonomy. For the schooling decision, for instance, it is likely that physical and decision-making autonomy are more important than emotional autonomy.

Our second main finding in this paper is that in AP and UP the school starting age is significantly affected by female autonomy levels, caste, religion characteristics, wealth and initial conditions. These results indicate that while children from some families start school at the prescribed age, others lose out. It is therefore not surprising

that there is a considerable starting school window as well as a substantial proportion of censored observations in these two states.

From the point of view of policy, our results therefore indicate that in states like AP and UP, where there is greater variability of SSAs, an attempt to target children in families with certain household characteristics would help to improve school outcomes. The importance of influencing schooling outcomes is borne out by the fact that the one variable that is significant for both states in influencing female autonomy as well as child education is the parent’s education. Results from both states therefore confirm that the impact on future generations will be greatest if we can educate current generations.

Furthermore, in these two states where many socio-economic characteristics influence the SSA, there are many policy levers that a government can pull to help improve outcomes. One of these is the autonomy of women. We note that this variable is highly significant in UP but less so in AP. Given that school entry is an issue in both UP and AP, our finding is encouraging for two reasons. First, the fact that one intrinsically valuable concept improves another can significantly increase the effectiveness of policies. By improving a woman’s position within the household, policy makers will not only improve women’s lives but will also contribute towards the second Millennium Development Goal of universal primary education. In addition to this, the findings of the present analysis provide policy makers with an additional tool for improving educational outcomes. Policies aimed at improving school enrolment should, therefore, not only focus on children but also on their mothers’ position within the household. The present analysis views female autonomy as an exogenous cultural factor but a number of studies try to endogenise this concept and investigate possible

determinants of a woman’s autonomy.51 The present results might be used to formulate effective policies for improving women’s situations, especially in areas where both female autonomy and children’s educational outcomes are low.