Public expenditure choices and gender quotas

This paper aims to be nested in the empirical literature examining the impact of gender quotas for elected posts to local government councils (panchayats) in India. Gender quotas apply at the level of both head (sarpanch, randomly assigned) and member (uniformly across councils). Received studies exploit the randomly allocated quota across panchayats at the level of sarpanch, and find a statistically significant impact of the gender of the sarpanch on public expenditure choices. This paper is motivated by the fact that those results imply sarpanch domination in the collective decisions of the council, and seeks to develop a model to show that such dominance is possible in the short run, but not inevitable. It then aims to test for sarpanch dominance using primary data from a field survey of panchayats in four states.

The model is tested on field survey data from a sample of 776 panchayats. The probit specifications test for factors explanatory of the choice of expenditure on waterworks as a binary variable, on the grounds that this is a smoother measure in multi‐year expenditure commitments. However, there are supplementary specifications testing for the quantum of expenditure on water, both as a share of the total, as well as in absolutes.

For the region surveyed, a higher probability of expenditure on waterworks is found in the presence of key variables that explain the incidence of water‐borne diseases like cholera and diarrhea, as ascertained from a separate set of specifications. The gender of the head is statistically insignificant. Thus, in the region studies, gender of the head is trumped by economic fundamentals in expenditure choices, but this leaves open the possibility that the (uniform) gender quotas at membership level may have been what aligned choices with fundamentals.

The key message of this paper is that the citizen candidate framework does not point to unique outcomes where public choice emerges from multi‐member councils. Following from this, any finding on the impact of a gender quota at the level of head will necessarily be context‐specific, and cannot become the basis for generalized expectations.