Forecasting coverage and non-revenue water with Markov process

This study utilises empirical data on service coverage (reflecting network expansion and new connections growth) and billing efficiency of National Water and Sewerage Corporation in Uganda (2003–2010). Markov processes were used to enhance the understanding and visualisation of explanatory factors underlying past trends as a basis for forecasting the future. Specifically, it was found that coverage (involving service up-take) and billing policies (affecting non-revenue water) may be adequate for operational purposes. However, policies can lead to a ‘less-than expected’ rate of exponential growth in the long-run, if attention is not given to associated uncertainties and explanatory factors. Sets of probability transition matrices, which present useful ‘intelligent’ forecasting properties that can be triggers for evaluating and designing infrastructure policy were derived. The study also highlights growth trends that are reminiscent of ‘absorbing’ states; such patterns characterise the achievement of policy objectives.