Small town water governance in developing countries: The uncertainty curse

Despite well meaning intentions, many aid interventions fail for one reason or another. The reasons are varied: lack of consideration of local circumstances and process requirements, and in particular inadequate involvement of affected stakeholders as well as inadequate cross-sectorial coordination. This is not surprising given poor organizational memory combined with decisions being made under time pressure and strict deadlines combined with little adaptive capacity. Additionally, information about the importance of process requirements and engagement is qualitative and as such is unfortunately often given secondary importance. To address this, we suggest a Risk assessment component as part of the project design phase based on Bayesian Networks (BNs) utilizing expert and local knowledge. This not only improves organizational memory and transparency but also provides a direct link for assessing cost benefits and minimizing the risk of failure. Most importantly this prioritizes engagement, processes and an understanding of the local context. This paper describes how BNs have been developed and tested on water supply interventions in the town of Tarawa, Kiribati. Models have been populated using data from interviews and literature to evaluate water supply options, i.e. rainwater harvesting, desalination and reserve extensions; this paper reports only on the model relating to reserves extension, i.e. new reserves for protection of groundwater extracted for water distribution purposes.