Did affirmative action work for Indian women?

India’s electoral affirmative action program, which assigns a random subset of districts to elect only female or low caste candidates, has been fodder for empirical researchers. The randomness, while rooted in principles of fairness, it also provides social experiments on a grand scale.

Raghabendra Chattopadhyay and Esther Duflo used it to show that female leaders invest in different types of public goods, like drinking water, that reduce work for women. They link up with Lori Beaman, Rohini Pande and Petia Topalova to show that exposure to women leaders doesn’t affect most people’s preference for male leaders, but it does reduce prejudicial attitudes towards women as leaders. That is, taste discrimination stays, but statistical discrimination goes down.

Today Rikhil Bhavnani is presenting a new paper here at Yale that looks at what happens when quotas randomly end. The answer? Constituencies previously reserved for women are five times as likely to elect a woman for office than ones that were never reserved for women.

Declining statistical discrimination might might be one reason that reservation effects persist. But Bhavnani argues that the main channel is a sort of incumbency and experience effect:  reservations introduce a cohort of women into politics who are able to hold onto party tickets.

This is good news, but does this mean that quotas only have a temporary effect?