Modelling pump functionality with a Markov process: insights and implications from Malawi McNichollDuncan ByrnsSydney 2018 This paper uses recent data on water point functionality from Salima District, Malawi, to predict the expected pump functionality rates using a model known as a Markov process. If the model fits, as the findings suggest, the implication for infrastructure sustainability is that long-term pump functionality rates will only improve if there is an increase in the probability that pumps will be repaired. Examples from Malawi, notably from Nkhotakota District, suggest possible methods for improving this probability of pump repairs through strengthening local stakeholder relationships, which may hold greater potential for improving infrastructure sustainability than the temporary benefits typified by direct project interventions.