Gender and Family Formation in Uttar Pradesh, India

Abstract

While modernising influences affect many facets of the lives of millions of Indian families, there remain deep-rooted socio-cultural practices and traditions that survive and become engendered in new institutional mechanisms. Labour market policy is but one example where age-old ethnic affiliations distort governmental efforts and find new ways of expressing themselves. Efforts over the past decade to slow the rate of population growth, by encouraging adoption of modern family planning methods have failed to tackle son preference and have caused the sex ratios at birth to be worse than at any other time in the nation's history. This is particularly so in urban India, even among the more educated populace, and it is worsening. This thesis sets out to assess the way in which such gender considerations affect family formation decisions, primarily concerning the quantity and quality of children, with an appreciation of the dynamic nature of the problem. First we assess how fertility preferences and past child outcomes affect the demand for family planning and how behaviours associated with the greater autonomy of women impact upon this process. The empirical work makes use of data from the latest round of the National Family Health Survey (NFHS) for India, 1999, for the state of Uttar Pradesh, in a simultaneous equation framework, in an effort to take account of the joint determination of many of the variables inherent in modelling such dynamic processes with cross-sectional data. We find that although women's autonomy has been held up as a means of achieving lower fertility, the two do not necessarily go hand in hand, unless coupled with the wider participation of women. Unless the primary social and economic motivations for preferring sons are tackled and dismantled through legislation and through changes to social attitudes, superficial policies to promote the well-being of women will have little real impact and may lead to worsening female child outcomes. One of the policies heralded to achieve the deeper goal of gender equality has been the promotion of education of female children, who as a group lag well behind their male counterparts on both literacy and numeracy rates. We thus turn our attention to investigating the way in which household time allocation decisions are made, focusing on the parental choice of each child's main activity; to go to school, to work in the home, or in the formal labour market, in an effort to understand how the household's opportunities and resource constraints, along with social norms impact such decisions. While some state governments are offering cash incentives to families to keep their female children in school and unmarried, significant labour market discrimination against women continues and constrains the value of this government investment. Making use of the same NFHS data for Uttar Pradesh, we estimate each child's trinomial time allocation with competing speci cations and then compare the results. The standard multinomial logit model is estimated initially but imposes some fairly tight assumptions on behaviour and the resultant data, that are unlikely to hold in the present application. A Mixed Logit model is then estimated that is able to bring greater flexibility and descriptive richness than is possible with the standard Logit model. Estimation results are compared and con firm the ability of the Mixed Logit to capture more fully the unobserved heterogeneity inherent in the data and to allow for correlation in the errors across children of the same family that is not permitted within the standard logit setup.