Search Climate Insights Directory

The document presents a method for analysing water-related poverty by integrating the Sustainable Livelihoods (SL) framework with Bayesian networks (BNs). This approach addresses the 'fuzziness' of livelihood systems—caused by variability, diversity, and uncertainty—by using probability distributions rather than fixed values. The authors introduce a novel way to represent institutions as 'categorical inequality' (e.g., gender) and a 'latent variable' method where indicators serve as indirect evidence for assets rather than direct proxies. The model was applied to data from Si Sa Ket province in Northeast Thailand, demonstrating that locally-originated water supply schemes, such as water tanks and drill wells, can be more effective for poor households than unreliable large-scale irrigation projects.