Experimental Design
The experiment uses a randomised encouragement design to identify the causal impact of parametric heat insurance. All eligible SEWA members can enrol, but grassroots leaders are incentivised to focus recruitment efforts only in randomly selected treatment villages (n = 2,815). Control villages (n = 2,816) receive no directed outreach. This generates exogenous variation in exposure to programme information and thus take-up.
We will estimate ITT and TOT effects using village-level treatment assignment as an instrument for individual participation. Additional analysis will use a regression discontinuity design around temperature thresholds that trigger payouts, exploiting sharp cutoffs in daily temperature records to estimate causal effects of receiving payouts.
We will test for heterogeneity by occupation, access to credit / savings, income, education, household composition, age, pre-existing health conditions, market and healthcare access, access to adaptive resources and treatment intensity (number of payments received).