Caste-Stratified Market Penetration Randomization
RCTClinical Trial
Measuring general equilibrium spillover effects requires cross-market variation in intervention coverage, but directly randomizing coverage rates at the village level can introduce selection bias or correlate with unobserved market characteristics. Stratifying individual-level randomization by naturally varying sub-groups (such as social castes) generates exogenous cross-market coverage variation while maintaining experimental control.
Picture this
Imagine testing how a new vaccine affects an entire town's economy depending on what percentage of the population gets vaccinated. Instead of picking some towns to receive 10% vaccination and others 90% (which might accidentally pick wealthier or healthier towns), researchers select specific local community clubs in every town and offer vaccines to members of those chosen clubs. Because some towns happen to have larger memberships in those selected clubs, the total percentage of vaccinated people in each town varies naturally across towns without introducing biased town selection.
What the evidence says
Generated exogenous cross-village coverage variation ranging from 0% to 53% for cultivators and 0% to 100% for landless laborers, with zero statistically significant correlation to village size, caste concentration, or number of castes after conditioning on eligibility shares.
- Who was studied
- 118 unique castes (jatis) across 63 sample villages in Andhra Pradesh, Uttar Pradesh, and Tamil Nadu, India.
- How
- Two-step experimental design randomizing 93 treatment castes and random sub-samples of households within castes to create market-level coverage variation.
What to do
Use sub-group or caste stratification when designing large-scale randomized evaluations to generate exogenous market-level treatment intensity without confounding market-level characteristics.
From the source
"We generate cross-village variation in these variables of interest through a two-step feature of our experimental design - we first randomly select a subset of castes (with varying population sizes) in each village to receive insurance marketing, and then offer random subsets of households within these castes rainfall insurance contracts."
Risk, Insurance and Wages in General Equilibrium