M-Maji works in three steps (see figure above for a system view):
Step 1: At the start of each day, water vendors use the M-Maji USSD (Unstructured Supplementary Service Data) shortcode to create a water advertisement consisting of their location, price, and purification method, if any. All advertisements are collected and stored in a central database and set to expire at midnight.
Step 2: Water buyers dial the M-Maji USSD shortcode to obtain a location-relevant listing of vendors with water, their locations, prices, qualities, and vendor ratings (see Step 3 for vendor ratings). Buyers can also access this information by sending in an SMS with the name of the village they are looking for water, at which point they receive back an SMS listing of all available water vendors in that village or—if water is not available in the selected village—the closest village with water.
Step 3: If a water buyer subsequently discovers that a vendor misreported water availability, price, or quality, the buyer can file a complaint with M-Maji via USSD. The database will keep track of complaints and alert future buyers of such negative histories through the use of vendor ratings.
Quality Control/SMS Alert System: Vendors that advertise “purified” water are monitored by M-Maji staff, who will conduct random water quality tests once a month to confirm the absence of fecal pathogens in the vendors’ water supply. Water sources that test positive for fecal contamination are removed as “purified” sources in the M-Maji listings, and an SMS alert system informs nearby users of water contamination.
M-Maji is designed to improve access to clean water by empowering residents with better information about water availability, price, and quality. By coordinating and centralizing water information from multiple sources, it provides to users information that might otherwise be unavailable. It also does so in a way that is economically sensitive, relying on basic GSM phones that are broadly accessible in slum communities and operating free of cost for users (USSD/SMS costs subsidized). Data accuracy is encouraged by the vendor rating system and the M-Maji support team on the ground, who will monitor the quality of our data through regular surveys and random evaluations (for example, through drop-in testing of water quality). Water sources that test positive for fecal contamination are removed as “purified” sources of water in the M-Maji listings, and an SMS alert system informs nearby users of water contamination.
M-Maji is a novel approach to water problems in Kibera. While alternative projects try to alleviate water problems by increasing water supply or enhancing its quality, M-Maji tackles the information side of the problem—the fact that communities struggle to locate and evaluate the quality of their water on a daily basis, and the possibility that this lack of information pushes the price of water to be higher than it should be based on supply and demand. Thus, by providing better water information to consumers, we might not only reduce the individual burden of finding clean water, but also put downward pressure on water prices, making clean water affordable and accessible to larger segments of Kibera’s population.
We believe that M-Maji will improve information about water availability, price, and quality, and that better information will: (1) reduce water search costs, in terms of time and money, (2) increase clean water uptake, and (3) encourage vendor accountability.
There is suggestive evidence in the academic literature that better water information may have these three effects. Randomized field experiments in developing countries, for example, have shown that households that receive information about water contamination are more likely to seek out a clean source and purify their water (Jalan and Somanathan 2008; Luoto 2009; Madajewicz et al. 2007). Research on ICT4D (Information and Communication Technology for Development) has illustrated that mobile technology can help improve market efficiency by reducing the cost of information (Aker 2010; Jensen 2007). And a large game theoretical literature indicates that opportunistic behavior may be curtailed by providing better information about people’s past behavior (Axelrod 1984; Greif 1994; Milgrom et al. 1990).