Elevating Women's Voices in Governance and Policy-making in India

2026-01-01

2027-07-31

Our primary outcome is the incorporation of women’s demands in local policy-making. Measuring the representation of demands in policy-making is complicated. We plan to do so using three methods. First, and primarily, we will leverage government administrative data on budgets, expenditures, and service delivery to compare actual policy-making across treatment arms. Second, we will conduct a mapping and audit of visible public services (see Chattopadhyay and Duflo, 2004). Third, we will use citizen reporting of preferences and access to services.

In addition to our primary outcome of improving the representation of women’s demands in policy-making, our theory of change outlines five additional secondary outcomes that we will measure, including: (1) women representatives’ political agency, (2) women representatives’ authority (as conferred/recognized by their community members), (3) women’s non-electoral political participation, (4) women’s electoral contestation, and (5) community gender norms and beliefs. These outcomes will largely be captured through surveys.

We will conduct a randomized control trial across 3,200 Gram Panchayats (GPs or local governments) and four states with our implementation partner, Transform Rural India (TRI). Our four study states – Chhattisgarh, Jharkhand, Madhya Pradesh, and Uttar Pradesh – were chosen because TRI already has substantial infrastructure to enable this study in these states, and we are feasibly able to implement our interventions in line with their election calendars. The next local government elections in two of these states (Jharkhand and Madhya Pradesh) are in 2027, allowing us to assess the impact of our interventions on women’s re-contestation and re-election, and with upcoming elections in Uttar Pradesh in early 2026, we will start our intervention following the election of new Sarpanch, allowing us to assess the impact of our interventions when they begin before norms of governance take root.

We will randomize three interventions: capacity and agency-building trainings to women elected chairpersons, the bi-weekly convening of peer networks of women elected chairpersons, and the quarterly convening of women-only representative-citizen public forums (Mahila Sabha). Within female-led GPs, we will have five treatment arms: control, capacity/agency-building trainings only, peer networks only, Mahila Sabha only, and TRI training + networks + Mahila Sabha. In male-led GPs, we will have two treatment arms: control and Mahila Sabha.

Not available

Randomization will be done using a computer program (e.g., Stata). Gram Panchayats (unit of randomization) will be block randomized to treatment arms, blocking on state, administrative block, caste reservation status, and levels of Self Help Group.

Randomization will be at the Gram Panchayat (GP) level. Our experiment will include 3,200 GPs (2,400 GPs led by female elected representatives and 800 GPs led by men).

No

No clustering.

Our primary unit of observation/randomization is the Gram Panchayat (GP). We will have 3,200 GPs total and will include data from the GP chairperson, their primary family member, and a mapping of public goods and services. We will additionally measure outcomes from citizens and plan to include 10 citizens per GP, for a total citizen sample size of 32,000.

The experiment includes a total of 3,200 Gram Panchayats (GPs). The allocation of these treatment arms is below.

Within female-led GPs:
Pure control (800 GPs)
Training only (400 GPs)
Peer networks only (400 GPs)
Mahila Sabha only (400 GPs)
Training + Networks + Mahila Sabha (400 GPs)

Within male-led GPs:
Pure control (400 GPs)
Mahila Sabha (400 GPs)