Experimental Design
We conducted the CDD program in 171 communities eligible for receiving the arsenic mitigation program. The mean community size is 153 households. We selected communities for inclusion based on their baseline level of arsenic contamination: (i) communities where more than 25% of sources of drinking water are arsenic-contaminated; (ii) or communities where arsenic-contaminated water sources are geographically concentrated. We randomly select 129 communities for receiving the program at public lotteries. Treated communities are further randomly assigned to three requirements for co-funding the installation costs: no contributions; cash contributions; labor contributions. Under the CDD program, communities first select locations for the subsidized safe sources of drinking water and agree on how to divide the community contributions, where applicable. These decisions are taken by consensus at a community meeting with representation of women and the poor, and in the presence of project staff who act as moderators. Communities must then raise or coordinate the community contributions before the wells are installed. Communities also take responsibility for maintenance and if necessary repair of the wells after they are installed.
After project implementation in treated communities, we carry out a field experiment where we elicit both stated and incentivized preferences over three different decision-making processes, each of which could be used to take decisions regarding installation of a future source of safe drinking water. The three decision-making processes are: decision-making in the perceived best interest of the community by implementing agency staff; decision-making using the community's own, pre-existing institutions; and consensus-based community decision-making, as in the original CDD program.
We will carry out the follow-up field experiment in all study communities in which installation of a new safe water source is expected to be hydro-geologically feasible. For communities where we know that installation of a safe water source is unlikely to be hydro-geologically feasible, it does not make sense to ask questions about a new, hypothetical source of safe drinking water. The expected sample size is 165 communities: 42 control communities and 123 communities that received the CDD program. Selectively dropping some treated communities potentially induces imbalance between treated and control groups, if the conditions which lead to failure in installation are not as good as randomly assigned, so we will report results with and without the area most highly affected by hydrogeological constraints to installation, which are geographically clustered.
The field experiment is an incentivized willingness to pay (WTP) elicitation. Specifically, we elicit participants' differential (WTP) for different institutional arrangements by offering a series of choices between two decision-making processes and two subsidy levels. For each pair of decision-making processes, we ask respondents to choose between one decision-making process at a baseline subsidy level, and the alternative decision-making process at a varying subsidy level. The baseline subsidy level corresponds to a community contribution of 3,000 Bangladeshi taka per installed tubewell. Choosing one decision-making process over another decision-making process paired with a lower (higher) subsidy rate implies a positive (negative) differential WTP for the first decision-making process. We will implement the WTP elicitation on average in 42 households per community, for a total of 6,905 participants. The study households in the main sample are randomly selected from administrative lists.
After administering the field experiment, we then assign 34 communities to receive a new water source by public lottery. For treated communities, this may be additional to the water source already offered under the original CDD program. For each community selected to receive a new water source, we determine the decision-making process and the subsidy level for the implementation as follows: first, we randomly select two out of the three decision-making approaches; second, we extract one pair of subsidy levels from those which were offered during the elicitation process; finally, we apply majority rule on the resultant choice between two combinations of decision-making approaches and subsidy level. This design makes the elicitation procedure incentive compatible and strategy-proof: each community has a positive probability of receiving the new source, and respondents should truthfully report their preferences, as long as they believe there is a non-zero probability that their preferences influence the final decision. The design also ensures that no choices expressed by any participant are revealed at any stage of the experiment, preventing confounding factors due to peer pressure or reputation concerns, and minimizing risks for individuals involved in the study.