People do not pursue education for its intrinsic worth alone but also for its potential to boost the earning and job prospects of individuals. If the labour market discriminates by rewarding women's education less well than men's education, the expected economic rate of return from educating girls will be lower than that from educating boys. Under such circumstances, even otherwise liberal-minded parents are likely to support education for daughters less than for sons.

Do these results imply that in India there is a weaker efficiency case for promoting education for girls? Not so, considering these estimates measure only private returns to education. Education of girls brings well-known social benefits such as reduced infant mortality and lower birth-rate. Though not quantified these dividends are thought to be substantial. Even so, the decision to enrol and retain a child in school is still essentially a parental decision in countries where primary education is not yet compulsory. In other words, the respective social benefits of schooling for girls vs. boys are not taken into account by parents when making decisions about the education of sons and daughters, whereas private returns are.

Would parents provide equal amounts of education to girls and boys if the expected private rate of return to girls' schooling was equal to or greater than that expected from boys' schooling? Societal norms such as early marriage and the dowry system would still militate against girls' education in South Asia. Moreover, in this and many other developing regions, where social security and state pensions are not generally provided, a male child will still give support to parents in their old age. But any benefits of educating females will be reaped by their in-laws. Such asymmetry in economic incentives for parents is probably a key barrier to equal treatment of sons and daughters, and it would be so even if returns from education for girls vis-à-vis boys were evenly matched. In terms of lessons for policymakers, these observations suggest there may be call for:

• public awareness campaigns to highlight the value of women's education. The message should be designed to shift conservative parental attitudes towards girls' schooling, on the grounds that many of the dividends of girls' education are non-economic benefits that accrue to the educated girl as well as to society at large, such as lower infant mortality rates
• steps by government to introduce legislation that bolsters economic incentives for educating girls, by reducing job and wage discrimination against women in the labour market. The signs are that if economic incentives for acquiring education are strong enough, cultural hurdles can be overcome.