Identities and public policies: Unexpected Effects of Political Reservations for Women in India

Reservations for women impact high-caste and low-caste women differently, decreasing the likelihood of a high-caste president. 

• When there are no reservations for women, high-caste men fill the majority of the Gram Panchayat seats, only 6.5% of presidents are women, and low-caste men are underrepresented.
• When there are reservations for women, 95.4% of presidents are women, and the majority of these women are low-caste.
• Reservations for women decrease the total number of candidates, which is entirely comprised of a decrease in high-caste candidates, reduces the probability that at least one high-caste individual runs for election (by 31 percentage points), and reduces the probability of electing a high-caste president (by 20 percentage points).

Reserving seats for women increased lower caste’s substantive representation through the implementation of policies preferred by lower castes.

• Reservations for women increase the provision of goods preferred by women. 
• Reservations for low-caste individuals cancel out the policy bias in favor of high-caste individuals. 
• Reservations for women cancel out the policy bias in favor of high-caste individuals.
• Reservations for women reservations do not only change policies towards the preferences of women, but also towards the preferences of low castes.

This research shows the importance of intersectionality in public policy implementation. In this case, a policy designed along one identity dimension (gender) alters the distribution of benefits along another one as well (caste). 
 

This study uses the 2005-2006 Rural Economic and Demographic Survey to gather nationally representative data of rural villages and households. Village-level Information of the survey includes what quotas exist, what characteristics the president holds, and which public works projects have taken place. Household-level information includes voting behavior and the public good preferences of the members of the household. This data included 270 Gram Panchayat elections in 168 villages, where 65 of these elections were for positions reserved for women and 97 were reserved for low-caste individuals. Two separate analyses of gender quotas occur in this study: The first analysis is of gender quotas and descriptive representation, and the second is of gender quotas and substantive representation.

For the descriptive representation test, a regression analysis is performed measuring the effects of gender and caste quotas on high-caste candidacy and presidency by each village, district, and election term.

For the substantive representation test, the public good preferences of individuals living in each district is calculated from the 2005-2006 Rural Economic and Demographic Survey responses. This sample consists of 1620 observations, as there are 270 districts with 6 public goods each measured. The public goods included in this study are water, sanitation, public lighting, communication, roads, and electricity. The public good preference for each district is measured across four groups: women, men, high-caste individuals, and low-caste individuals. This figure indicates the percentage of a group who wanted the Gram Panchayat to allocate money for the public good in question. Preference differences by gender and caste are calculated by subtracting the preference rates of men from those of women and the preference rates of high-caste individuals from low-caste individuals. Another regression is performed to measure the impacts of reservations for women and low-caste individuals, as well as the public good preference difference by gender and caste, on the provision of the public good in question in a set village, district, and election term.

Robustness checks adding wealth and education as variables were performed.