NYT Magazine has an excellent article by Tina Rosenberg, highlighting the problem of discrimination against the girl child and the "daughter deficit" (or "missing girls"), widespread in many poor societies where having a son is seen as a financial and social necessity. The article is an excellent example of how conventional and easier to implement best-practice prescriptions of arm-chair policy makers under-estimates the complex challenge of addressing social issues and thereby fails to achieve the desired results.
I am inclined to believe that such complex socio-economic problems provide fertile ground for micro-economists and behavioural scientists to design public policies that would "incentivize" or "nudge" parents towards girl children (or away from a son preference). Such attempts have to be rooted in the specific local social and economic context and may need to go against the grain of political correctness and unsettle some of the entrenched traditional conventions, customs and institutions. Conventional regulatory and trickle-down development approaches may need to be reinforced with these subtle and context-specific nudges (like nudges to stigmatize dowry taking) and incentives (say, tax-breaks for inter-caste marriages) if we are to expedite the process of desired change on these issues.
Conventional wisdom has it that a four-pronged strategy of enforcement of strong regulations (banning foetal sex identification, dowries etc), awareness creation, economic development ("the wealthier the home, the more educated the parents, the more plugged in to the modern economy, the more a family will invest in its girls"), and aid to women ("a mother who has more money, knowledge and authority in the family will direct her resources toward all her children’s health and education") will bring about an end to the selective discrimination against girl children. But, as the article points out, there is ample evidence to suggest that such optimism glosses over the sheer complexity of the challenge.
Researchers have found that "girls are actually more likely to be missing in richer areas than in poorer ones, and in cities than in rural areas", "having more money, a better education and (in India) belonging to a higher caste all raise the probability that a family will discriminate against its daughters", and "while increasing women’s decision-making power reduced discrimination against girls in some parts of South Asia, it made things worse in the north and west of India". Fascinatingly, it has been found that the level of discrimination varies between the first and subsequent girl child, with the first daughter treated like her brothers. However, "a subsequent daughter born to an educated mother was 2.36 times as likely to die before her fifth birthday as her siblings were to die before theirs — mainly because she was less likely to see a doctor".
Fundamentally, the sociological "son preference" has perceived economic underpinnings - male children earn, they look after parents, they require dowries etc - which can be addressed only by increasing the "economic returns to girls". The "deeply embedded son preference associated with highly patriarchal social systems" means that all conventional policies have to be supplemented with civic mobilization and social movements to "change customs regarding marriage and inheritance associated with patriarchal kinship systems, which favor males".
Difficult to define policy actions for these and even more difficult to implement and achieve the desired outcomes.
See also very good articles on discrimination against women here and here.