A Micro-Based Evaluation of  Water Projects

Raymond Boadi Frempong, David Stadelmann. Show Author details

Evaluation Water Projects
Water flowing from a tap in Senegal.

A large proportion of the world’s population (about 884 million people according to WHO & UNICEF’s 2017 estimates) lack access to clean potable water. More than half of the affected people live in Sub-Saharan Africa and other developing regions. To remedy this, there have been many development efforts in the water and sanitation sector to provide water infrastructure to improve the conditions in these regions. Such projects could offer welfare gains, including reducing water-borne diseases and water collection times by women and children. Given that development and research funds for these projects are limited, it is important that researchers find cost-effective means to evaluate these projects. Our study (see here) explores the potential of using existing secondary data sets to credibly assess the impact of many improved water provision projects on households’ welfare indicators such as water usage, health and water collection time.

The evaluation of water projects as an academic exercise is not new in the literature. However, several methodological problems in the existing studies can systematically affect the validity and generalizability of their results. For instance, the cost of randomisation often hinders the large-scale evaluation of water infrastructure projects. As a result, research projects often use small sample sizes with limited geographical and temporal scope. 

We suggest a substantially less costly approach to evaluating water provision projects. We match existing water supply programme data with geo-coded household panel surveys. Apart from very high cost-effectiveness, our approach has other significant advantages: (1) it allows us to study the impact of water supply investments on a large, nationally-representative sample over a longer period; (2) when panel data is available, the approach allows for controlling unobservable time-invariant factors; (3) given the richness of most household surveys, it is possible to control for many confounding time-variant factors; (4) it can lead to a reasonably credible causal identification of the effect of improved water provision where randomised control trials are not feasible or too costly; (5) due to their low frequency and the variety of topics addressed in household surveys, estimates of treatment effects are less likely to suffer from a Hawthorne effect (i.e., a bias that occurs if the act of being surveyed itself causes behavioural change).

Improved water supply in Uganda

We apply our approach to investigate the effect of water supply on household welfare from 2009 to 2012 in Uganda. In particular, we calculate the improved water supply capacity for sub-counties in Uganda by relating each improved water source with the number of people it can serve. We then divide the total number by the sub-county population size to obtain the improved water supply rate. This indicator accounts for the different capacities of the various water projects and sources in the sub-county. The data for this calculation is obtained from the Uganda Water Supply Atlas (see here). The classification of improved community water sources closely follows the definition of the Joint Monitoring Program for Water Supply, Sanitation and Hygiene (see here).

Our quantitative analyses show the impact of improved water supply on five individual welfare indicators in Uganda. The first is improved water usage which indicates whether the household’s drinking water is from an improved water source. The second and refined indicator of improved water usage is directly based on the Joint Monitoring Program for Water Supply, Sanitation and Hygiene. With five ordered outcomes, it considers the water quality and total water collection time (including queuing time). The third indicates an acute water-related illness (such as diarrhoea and vomiting) in the household in the last 30 days. In keeping with the development literature on water and household time use (Boone, Glick, & Sahn, 2011; Gross, Günther, & Schipper, 2018), we provide analyses of the effect of improved water provision on time spent on water collection and distance to the primary water source. These household-level measures are obtained from a geo-coded household survey from the Uganda National Panel Survey between 2009 and 2012 (see here). We match each household to the corresponding sub-county improved water supply rate from the Uganda Water Supply Atlas using the geographical locations. Finally, we estimate the effects of improved water supply on the household’s use of improved water, the incidence of water-related symptoms, water collection time and distance to water sources.

The impact of access to clean water

Our results show that closeness to improved water installations is associated with a higher chance of using water improved water. Increasing the improved water supply rate by ten percent is associated with about five increases in the probability that households will use water from these sources. We also noticed that women and children travel shorter distances to collect water. Reported distance to water decreases by about 19 km for a unit increase in the improved water supply rate. However, we find no significant effect of improved water supply on the incidence of diarrhoea.

The contribution of our study, in addition to empirical findings, is the suggestion of a scalable evaluation approach. The study employs available datasets but merges them for impact evaluations, by combining administrative information on water supply with household socio-economic panel data. Methodologically, the approach allows us to employ a more sophisticated quantitative analysis that considers household and individual characteristics that do not change over time when investigating the potential impacts of development projects. Thus, our approach enables us to credibly identify the possible effects of development projects.

Further information:

  • This article is based on the following publication by the authors: Raymond Boadi Frempong, Lucas Kitzmüller, David Stadelmann (2021). A micro-based approach to evaluate the effect of water supply on health in Uganda. World Development, 145. (Access it here)
  • The associated image of this post was taken from Wikimedia Commons (link).
  • The authors gratefully acknowledge funding from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy – EXC 2052/1 – 390713894.


  • Christopher Boone, Peter Glick & David E. Sahn (2011). Household Water Supply Choice and Time Allocated to Water Collection: Evidence from Madagascar, The Journal of Development Studies, 47 (12), 1826-1850.
  • Elena Gross, Isabel Günther, and Youdi Schipper (2018). Women Are Walking and Waiting for Water: The Time Value of Public Water Supply. Economic Development and Cultural Change, 66 (3), 489-517.

Author details

Postdoc (University of Bayreuth)
Visit personal website

Professor (University of Bayreuth)
Visit personal website