Raj Arunachalam and Trevor Logan conclude in an NBER working paper that there is no evidence to prove that "real dowry amounts have systematically increased over time in South Asia".
Whatever be the sociological reasons, and there are a number of them, from an economic point of view, the major incentive that drove the dowry market was an economy in which women were not formal participants. The increasing role of women in workplaces should by itself ensure a significant drop in the dowry rates.
At the risk of being politically incorrect, the economic rationale for this decrease would be that the Net Present Value (NPV) of the future revenue streams of all literate and employed women is most often likely to be much more than any one time marriage pay-off. It may also be true that male literacy may have an attenuating effect on dowry amounts, though I am inclined to beleive that this may be a small effect!
In fact, it may be a good idea to have studies that control for certain parameters, so as to bring out trends, if any, indicating this aforementioned hypothesis. Such studies should involve time series data on dowry trends across categories like
1. States where women's literacy and employment rates have been increasing with those where these rates have been relatively stagnant.
2. States where men's literacy rates (here more focussed search could go on to secondary and higher education figures too) have been increasing with those where it has been stangnant (this addresses the demand side dimension)
3. Similar studies can also be carried out across certain castes/communities where the propensity to give dowries is especially high.
4. Proportion of women who are employed but have not given dowry.
5. Average dowries given by educated (and employed) and uneducated (and unemployed) women.
6. Average dowries received or taken by educated and uneducated men.
7. Differences in the dowries paid by the differently employed (or unemployed) daughters of the same family. It is logical that the employed daughter pay (or not pay) a much smaller dowry than the unemployed.
8. This analysis can be extended to cover all the married daughters of an extended family, so as to control for other sociological foctors.
9. Dowries given can be traced against the incomes of employed women, controlling for parameters like community, regions/states, and socio-economic backgrounds.
I am more inclined to believe that social empowerment of women, by itself, will not be enough to significantly reduce or eliminate dowries. More so in societies like ours with its deeply entrenched customs and social mores. If social empowerment were enough, then the average dowries should have decreased dramatically in states like Kerala and among the Self Help Group (SHG) members of Andhra Pradesh and Kerala. But there is little indication that this has happened.
It would therefore appear that economic empowerment of women is a sine qua non for ensuring that dowries become a thing of the past.