International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Genius Murwirapachena

Last modified: 28 March 2017



Genius Murwirapachena[1] and Johane Dikgang[2]



The best way to develop consumer oriented water policies is through assessing whether current water service packages provided to households meet their expectations. In South Africa, the provision of water services to household is a direct mandate of municipalities. The South African Constitution considers access to water as a basic human right, which makes water both a social and economic good. In light of this, South African municipalities have a mandate to provide water services at tariffs affordable to all citizens. The increasing block tariff (IBT) water pricing system where higher tariffs are charged to households consuming higher quantities of water is common among municipalities. Since access to water is a basic human right, households that cannot afford to pay for water services are entitled to receive at least 6000 litres as free basic water per month.

Most municipalities in South Africa struggle to raise enough revenue to cover the costs of providing water services, thereby recording budget deficits each successive year. Some critics suggest that municipalities fail to generate enough revenue from water services because of inefficiencies while others suggest that municipalities do not properly consult water consumers on their preferences. In most areas, households receive “one-size-fit-all” water services and do not receive the value of their money. On the other hand, some scholars blame backlogs created by the exclusive policies of the apartheid era. Considering these arguments and the huge financial losses recorded by municipalities, the question that comes to mind is, how best can municipalities improve on water services provision? Improving on water services provision is not only essential for revenue raising but is also useful in sustainable water resources management. South Africa is a water stressed country and the available water resources should be sustainably managed. Since households play a key role in water consumption, it is essential to establish how they prefer to receive water services. The main object of this study is to establish households’ preferences for water services.

Choice experiments (CE) were used to elicit households’ preferences for water service packages. The study was conducted in the eThekwini Metropolitan Municipality, located in eastern South Africa, in the KwaZulu Natal province. This municipality was chosen because it has both an urban and a rural component which capture the dynamic and diverse structure of South Africa. Attributes used in the choice experiments are access to piped water, source of water, reliability of water supply, water pressure, water quality, and total cost of water per month. South African cities typically consist of suburban and township areas. The former are higher income areas while the latter are low income areas. Therefore, the samples in these two areas are distinct, implying different status-quos. As a result, this study stratifies the sample into two strata, one for suburb households and the other for non-suburb households. Two distinct questionnaires were developed for each stratum as each has a different status-quo. The study also tests for the relevance of including the status-quo alternative in choice experiments by providing an additional questionnaire with an inapplicable status-quo to each stratum and then compares the results to those with the most applicable status-quo.

In designing choice sets, most studies in literature use orthogonal design and most recently efficient designs. Efficient designs are only efficient if the prior parameters are correct. However, if the priors are wrong, an efficient design can become inefficient. Recent studies suggest that the Bayesian D-efficient design is more robust because its efficiency is less sensitive to misspecification of the priors. Therefore, this study uses the Bayesian D-efficient to design choice sets.

A total of 1000 households from the eThekwini metropolitan municipality were surveyed (500 households from suburbs and another 500 from townships, informal settlements and rural areas). For each stratum 250 responses where collected using the questionnaire with the applicable status-quo and the other 250 using the questionnaire with the inapplicable status-quo. Households’ responses were analysed using the conditional logit and multinomial logit models. Firstly, suburban responses collected using the questionnaire with the applicable status-quo were analysed and compared to responses from the same stratum but collected using the questionnaire with the inapplicable status-quo. This was done to establish whether the inclusion of the status-quo alternative is critically essential in choice experiments. Secondly, responses collected using the questionnaire with the most applicable status-quo from non-suburban areas were analysed and compared to responses from the same stratum but collected using the questionnaire with the inapplicable status-quo. Thirdly, since the same attributes and levels were used in the questionnaires for both strata, responses collected using the questionnaires with the applicable status-quos were combined, analysed and results compared to those obtained in the first and second steps. Finally, responses for the two strata collected using the questionnaires with the inapplicable status-quos were also combined and analysed.

Preliminary findings from the study revealed that estimation results for responses collected with questionnaires that had inapplicable status-quos do not significantly differ from those for responses collected using questionnaires with applicable status-quos in both strata. It was also found that households from the suburbs value the quality of water and its reliability more than the other attributes. As expected, households from townships, informal settlements and rural areas are more concerned with the monthly cost of water services. The latter would also prefer to pay for piped water inside the dwelling and are not so much concerned with water reliability and pressure. The policy implication for this is that the municipality should maintain the provision of reliable good quality water in suburban areas. For households in township, informal settlement and rural areas, water inside the dwelling is very important.


Keywords: choice experiments, suburban, township, water


[1] School of Economics and Development Studies, Nelson Mandela Metropolitan University, Port Elizabeth, South Africa. Email:

[2] Department of Economics and Econometrics, University of Johannesburg, Johannesburg, South Africa. Email:

Conference registration is required in order to view papers.