Chlorine dispensers in Kenya: scaling for results, 3ie Grantee Final Report, 2015.
Authors: Amrita Ahuja, et al.
We conducted three studies: one survey experiment and two large-scale randomized evaluations, to investigate how a particular community-level water treatment device, the chlorine dispenser, is valued and allocated by local government officials, and how best it can be financed and managed.
The first study involved 179 elected county councilors in rural Kenya, who chose between different public goods packages in an incentivized survey. The public goods varied with respect to two attributes: how the location of the infrastructure was decided and who controlled the funding associated with maintaining it. We found that on average, councilors valued the opportunity to target the water treatment technology, but not the ability to control funding for maintenance.
The second study, Community Financed Dispensers (CFD) concerned how to finance the chlorine refills required to keep chlorine dispensers functional. The chlorine dispenser is a classic common good in that chlorine is rival in consumption, yet it is not practical to restrict access to it. 104 communities were randomly assigned to a free provision arm or one of four community financing arms. In the community financing arms, water source users were responsible for purchasing chlorine refills. In half of these, an upfront contribution to the cost of the dispenser was required prior to installation (“up-front payment” treatment); cross-cut with this was a “threat of removal” treatment, in which the dispenser was a portable model that could be removed if the community failed to stock it with chlorine. In free provision communities, permanent dispensers were installed and chlorine refills were provided at no cost. We found no effect of up-front payment on household chlorine usage three months after installation, nor on whether the dispenser contained chlorine at unannounced visits over a period of eleven months. In contrast, threat of removal led to a 20 percentage point increase in the probability that a dispenser was stocked relative to permanent community-financed dispensers, which were stocked 41% of the time. Further, we estimate that in communities receiving free refills, enough doses were consumed to consistently treat all households’ drinking water. Households in threat of removal communities consumed 28-30% of the
recommended doses and households in other communities consumed 20% of the recommended doses.
