Nearly one-fifth of the world's population lack access to improved sources of drinking water. Diarrheal disease causes nearly 1.8 million deaths worldwide each year, and is responsible for 17% of deaths of all children under five years of age. Poor water quality also contributes to other debilitating diseases such as schistosomiasis, trachoma and worms. This problem is acute in rural Africa, where over half the population lacks access to improved water.
While there is consensus on the severity of the problem, there is little robust evidence on the effectiveness of solutions relevant to the rural poor. Piped treated water, which produced substantial health gains in middle-income and rich countries, is not considered feasible in rural areas of poor countries with dispersed populations and weak institutions. Community interventions short of piped water, such as spring improvement or communal wells, have not produced strong results. As a result, policy-makers are increasingly interested in point-of-use treatments, such as chlorination and filtration. In this context, accurate measurement of household willingness-to-pay (WTP) for clean water technology is crucial for policy. Knowledge of this parameter can guide decisions on the magnitude and targeting of subsidies for household-level technologies, and help quantify the benefits of community level, infrastructure-based solutions. However, robust measurement of WTP is difficult. Observational studies may suffer from selection bias; contingent valuation methods may be biased or imprecise if respondents answer strategically or without careful thought.
This study tests the efficacy of the Becker-Degroot-Marschak (BDM) mechanism, a valuation elicitation technique from experimental economics, in measuring demand for a household ceramic water filter. BDM is incentive compatible; it is in the respondents’ best interests to state their willingness to pay truthfully. Furthermore, it provides an exact valuation, in comparison to a more traditional take-it-or-leave-it offer at a randomized price (TIOLI), which provides only an upper or lower bound. The mechanism has other benefits, including that it generates random variation in terms of which respondents receive the filter and what price they pay.
To test whether BDM provides reasonable results in practice, we offer a ceramic water filter for sale to 1600 rural households in Northern Ghana. Households are randomly assigned to BDM or to TIOLI treatments, and we compare the resulting estimates of demand. We also incorporate slight variations on both BDM and TIOLI, including changing the agent of randomization and informing participants of the market price of the filter, to see whether these variations improve or diminish the relative performance of BDM.
Despite its appealing theoretical properties, BDM has not been rigorously evaluated in a developing country context. If BDM can be shown to work well in the field, it could provide a useful tool for estimating demand for a wide variety of products.
