Some societies have preference over boys. Perhaps because boys will carry the family's name, but also because they can rely on their boys to do heavy works. Other societies prefer to have girls; because in their cultures girls are responsible to take care of their parents in the old days. Or, parents may not have any preference over a certain gender, but they want to have both genders in the family. Each preference will have implications in the size and resource allocation of the family.
One implication of gender preference is gender discrimination - in terms of within-household resource allocation. Deaton (1989) measured the boy-girl discrimination in terms of the 'reduction of household expenditure on adult goods.' (Adult goods: tobacco, adult clothing, alcohol, eating out, etc.). When a kid is valued more, parents will be more willing to reduce their spending on adult goods. He found no evidence of gender discrimination in Cote d'Ivoire, and a small and insignificant bias in favor of boys in Thailand.
We can also measure boy-girl discrimination in terms of health outcome. Using data from rural Punjab, India, Monica Das Gupta (1987) found that the mortality rate of children below one month is higher for boys. However, as the children gets older, girls mortality rate surpasses boys. Child mortality rate is also much higher for girls who were born as the second child and over.
The fact that boys have higher probability to die within a month after birth is quite logical. Boys have the XY chromosome, while girls have XX. Remember that the Y chromosome is a 'mutated' version of the X - means that boys are by nature mutants (the 'defect' version of girls). That makes boys are more prone to death, which explains the higher rate of postnatal mortality rate. The higher rate of girl mortality rate at the older age reflects the different treatment of parents. For example, girls receive lower nutrition, less clothing etc. And the value of girls is even less when parents have already had more survived children.
My professor at Harvard, Robert Jensen (2006) raised another issue. Parents may not necessarily discriminate against girls (or boys). But they may still prefer a certain gender (let's say, boy). When the first child is a girl, parents is more likely to have another one. When the second one is still a girl, it is more likely for them to have a third one, and so forth. As the result, girls tend to come from big families. Even though parents don't discriminate, coming from big families, girls will have smaller allocation of household resources ("equal treatment, unequal outcome").
How true is that? The answer can't be theoretical - it should be empirical. That may also depend on where do we do the research. But according to his preliminary finding, some Indian states where preference over boys are strong tend to confirm this results.
Childrenomics | Gender