Feasibility of financing nature-based solutions for water security through water tariffs: Evidence from South Africa

Healthy water catchment areas regulate the timing, quantity, and quality of stream flows, saving on grey infrastructure costs (Rebelo et al. 2021). As such, natural water catchment areas constitute a form of ‘ecological infrastructure’ (Adamowicz et al. 2019). However, in South Africa, the spread of invasive woody plants and poor farming practices have degraded many water source areas, reducing their capacity to deliver such ecosystem services (Nesshöver et al. 2017), impacting on water security. Studies have shown that the restoration of this ‘ecological infrastructure’ could be a cost-effective and environmentally sustainable strategy to improve water security (Vörösmarty et al. 2021), especially when one considers the broader co-benefits to society, such as recreation and tourism opportunities, protecting grey infrastructure, and buffering communities from the impacts of climate change (Jones et al. 2012; Logar et al. 2019; Choi et al. 2021; Rasmussen et al. 2021).

A significant obstacle to the success of catchment restoration and conservation projects, particularly in the Global South, is funding (Shackleton et al. 2017). Notwithstanding its economic justification, catchment restoration is expensive, both initially (e.g., planning and project design costs, construction costs, acquisition costs) and in the longer term (i.e., annual maintenance costs). In many countries, the management of catchment areas is largely dependent on state budgets, which are insufficient for securing their long-term health. Furthermore, state investment in ecological infrastructure is often deterred by perceptions of high risks and uncertainty of returns as well as the long timeframes involved relative to grey infrastructure solutions. While several robust business cases have culminated in the establishment of initiatives to invest in catchment conservation such as water funds and public–private partnerships (TNC 2018), leveraging sustainable funding streams for such initiatives remains a significant challenge, especially in the Global South (Mbopha et al. 2021).

However, there may be significant untapped potential for generating finance for catchment restoration and conservation through revenue generated from water services (Rogers et al. 2002; Makwinja et al. 2019; Mu et al. 2019). In particular, water tariffs should be considered the main source of finance for catchment restoration and conservation because water supply is a service, and consumers are generally willing to pay more for efficient, high-quality services (Grafton et al. 2020). A recent meta-analysis found that in the Global South, where quality service delivery is less likely to be taken for granted, households were willing to pay three times more for water than Global North countries relative to their income (Roldán et al. 2021). However, in most countries, water prices are kept low to minimise the social burden and ensure access (Majumdar & Gupta 2009). Consequently, water is often priced far below the cost of supplying it, resulting in economic inefficiencies, wasteful use, and infrastructural decay (Grafton et al. 2020). If designed in a way that is affordable to the user while also being financially sustainable for the supplier, water pricing structures can be an effective instrument in achieving environmental, social, and economic goals.

Water pricing has both incentive and revenue effects. Increasing the price of water incentivises consumers to reduce water consumption and invest in water-saving technologies (Mu et al. 2019). Where water demand is price-elastic, this may reduce revenues, but if it is inelastic, then increased tariffs could generate additional revenues that can be used to finance water resource management, including the restoration and conservation of water source areas. Nauges & Whittington (2009) found that own-price elasticity of water in developing countries tended to be inelastic, typically ranging from −0.3 to −0.6. In South Africa, Hoffman & du Plessis (2013) also found that the demand was relatively price-inelastic, with higher prices increasing revenues more than they reduced consumption. Therefore, increasing water triffs are likely to increase revenues with water saving being a secondary benefit.

Numerous studies have investigated the opportunities to mobilise financial resources for the investment into water-related ecological infrastructure within the public and private sectors (Bennett et al. 2014; Gómez-Baggethun & Muradian, 2015; Mbopha et al. 2021). Other studies have recommended water tariffs as a potential funding mechanism for catchment conservation (Colvin et al. 2015; Cartwright 2021). However, few, if any, studies have established whether the additional revenue generated from aggregate household willingness to pay (WTP) for water would be sufficient to cover the costs needed to fund catchment restoration.

This study investigates the extent to which water tariffs for domestic water supply could be increased to raise revenues for catchment restoration in three coastal metropolitan municipalities in South Africa. The objectives were to (i) determine households' WTP for their existing water use, (ii) investigate whether institutional trust and municipal satisfaction influenced WTP, and (iii) establish whether aggregate consumer surplus at the municipal scale could cover the costs needed to finance catchment restoration. This study is particularly pertinent given that most coastal cities in South Africa are expected to face more frequent water shortages in the future due to climate change.

The study was carried out in three of the eight largest metropolitan municipalities in South Africa: the City of Cape Town, Nelson Mandela Bay (which includes the city of Gqeberha), and eThekwini (which includes the city of Durban). These municipalities all face growing demands for water from catchment areas that are becoming increasingly degraded. Accordingly, water funds have already been established or are in the process of being established to address catchment conservation for all three. These aim to capitalise funds for catchment restoration from a mixture of public and private sector sources (TNC 2018; Cartwright 2021). In 2016, the combined population of the three municipalities was 8.3 million people, which is approximately ∼16% of the total South African population (Small 2017).

The City of Cape Town relies heavily on surface water, which it receives from three important water source areas: the Boland Mountains, the Groot Winterhoek, and Table Mountain. Of these three areas, the Boland Mountains are responsible for supplying the City of Cape Town with 97.1% of its water. These three important water source areas span over 1,700 km² and drain into five major dams.

Residents in the Nelson Mandela Bay municipality receive water from the Algoa Water Supply System, which is subdivided into Western, Central, and Eastern systems. The Western System supplies water to approximately 70% of the residents in the Nelson Mandela Bay municipality and is made up of three catchment areas: the Baviaanskloof, Kromme, and Kouga, which collectively cover 5,610 km². The Central and the Eastern systems supplement the Western System through a system of dams, transfer schemes, and springs.

The major cities of Durban and Pietermaritzburg receive most of their water from the Greater uMngeni catchment, a small but significant catchment, which covers 4,349 km². The available yield of the catchment is not enough to meet the demand for water; thus, the catchment has been fully developed to accommodate this demand, with four large storage dams constructed over the years.

The survey questionnaire was divided into four sections: (i) household characteristics, (ii) WTP for existing water use, (iii) trust in municipality, and finally (iv) socio-economic status (see questionnaire in Supplementary Table S1). The first section covered general information about the respondents' household such as their household size, and whether their property had a garden, swimming pool, borehole, wellpoint, rainwater tank, and/or pumped greywater system. Respondents were also asked about their average household monthly water bills and how they pay their bills. Respondents were asked how confident they felt about their reported utility bills on a five-point Likert scale from ‘not at all confident’ to ‘extremely confident’, since they were not able to access their utility bills while completing the questionnaire.

In the second section, WTP for existing water use was elicited by means of the contingent valuation method (Carson 2000). Following Makwinja et al. (2019), a double-bounded dichotomous choice format was used, which is a ‘take it or leave it’ approach in which a respondent is asked to respond yes or no to a proposed situation or payment, and a follow-up question is asked with a slightly altered payment option to reduce the variance in WTP estimates (Van Song et al. 2019). If the respondent answers positively to the first bid, the second amount offered is higher. However, if the respondent rejects the initial bid, the second amount offered is lower. This is considered a more statistically efficient approach than a single-bounded dichotomous choice question (Carson 2000).

In most contingent valuation studies, bid amounts are finalised prior to surveying and are generated randomly. However, this may lead to starting point bias (Carson 2000). This study aimed to mitigate starting bias by offering an initial bid that was 1.5 times respondents' reported water bill. This is a similar approach to Makaudze (2016) who used the average water bill of the study area as the initial bid. If the response to the initial bid was ‘yes’, the subsequent bid offered was double the respondents' water bill. However, if the response to the initial bid was ‘no’, the subsequent bid was reduced to 1.25 times their reported water bill. Respondents who could not estimate their water bill were removed from the analysis. Similar to Peletz et al. (2020), dichotomous choice questions were followed with an open-ended question asking respondents to state their maximum WTP for water. This offered an opportunity to check whether respondents' answers were consistent across both questions.

The third section included questions about the respondents' degree of trust and satisfaction in the municipality. Respondents were asked to indicate their level of satisfaction with the way the municipality manages water supply and sanitation in their neighbourhood on a five-point Likert scale, from ‘very dissatisfied’ to ‘very satisfied’. Respondents were also asked whether they think the municipality should be charging more for water and whether they would be able to trust that their municipality would use the funds to improve water security if they increased the water price.

The final section dealt with the socio-economic background of respondents. Data were collected on income, race, age, and education. These factors were considered to validate the responses.

Data for this study were collected over a period of three weeks from household residents in the three municipalities. Enumerators were trained to capture participants' answers on KoboCollect, an open-source application (KoBoToolbox 2019). Prior to the interviews, the questionnaire was pre-tested to improve the flow of questions, to reword ambiguous questions, and to exclude any variables that may be irrelevant (Majumdar & Gupta 2009). Interviews were conducted in waiting queues outside of the main offices of the Department of Home Affairs (DHA) in the home language of each participant (English, Afrikaans, Xhosa, or Zulu). The DHA's core functions are to issue identification and travel documents; to manage birth, death, and marriage certificates; and to grant citizenship and residency permits. Due to the range of identification services that the DHA offers to the citizens of South Africa, these queues are frequented by a wide representation of household residents. While this is a form of convenience sampling, it is the authors' experience that door-to-door surveys in South Africa tend to have a strong bias away from wealthier households due to security issues leading to high levels of refusal.

For the statistical analysis of WTP, the dataset was cleaned to remove all outliers, those who had provided ‘prefer not to say’ answers for income-related questions and those who did not have an income. The outliers were identified by assessing the value that was reported by respondents as their monthly water bill. This value was deemed outside (higher) than what was expected based on the following factors: how the water bill compared to the stated electricity bill; the respondent's income category; how the income category related to the stated water bill; the suburb in which the respondent lived; the number of people in the household; and whether the stated water bill, when high, corresponded to having a garden or swimming pool.

This resulted in a dataset with seven explanatory variables. The function ‘dbchoice’ in the ‘DCchoice’ package of R 4.0.2 was used to carry out the analysis of the study (R Core Team 2022). This package employs the generalised linear model function, which contains a binomial logit argument to explore the association between WTP and explanatory variables (Table 1). Bootstrapping was also used to estimate the confidence intervals for the estimates of WTP (Krinsky & Robb 1986). For each of the four WTP estimates, this method generated a lower and upper bound of the interval. To validate the results, the data were examined in terms of variables including municipality, household income, gender, age, satisfaction with municipality, trust in municipality, and highest educational level of the respondent.

To calculate aggregate WTP per municipality per annum, average WTP per income category was multiplied by the total number of households that receive piped water to their house (as per the national census), which was then summed. Aggregate water bill per annum was calculated using the same method, and the difference between these two values was used to calculate consumer surplus. Only households earning above R6,400 per month were considered in these calculations, as households below this threshold were considered indigent (i.e., able to receive free water). This paper therefore assumes that households below this threshold are unable to pay for water (even though some households may not have applied for the indigent support programme).

Among the 1,959 individuals who were approached, 1,500 agreed to complete the interview (76.6% response rate). Of these, only 502 individuals met the interview criteria. The criteria for acceptance were that respondents (i) were over 18 years old, (ii) lived in the surveyed municipality, (iii) considered themselves a financial contributor or decision-maker in their household, (iv) received piped municipal water to their house, (v) paid some amount towards their water bill, and (vi) could provide their household monthly water bill.

The average age of respondents was 41, and almost half of the respondents were between the ages of 18 and 39 (Table 2). Respondents came from a total of 171 suburbs, and the mean household size of respondents was 4.89 (±2.1, range 1–19). About 36.4% passed their final year of high school (matric), and 46.2% had a tertiary level qualification (diploma or degree). Average household monthly income was R39,393 in eThekwini, R35,819 in the City of Cape Town, and R15,646 in Nelson Mandela Bay.

There were significant differences in the proportion of households per income category in all three municipalities compared to that of the most recent census (Statistics South Africa 2012). The sample from eThekwini differed the most (X² = 94.8, p < 0.001), followed by the City of Cape Town (X² = 55.0, p < 0.001) and then Nelson Mandela Bay (X² = 22.3, p > 0.01). When the highest income category (R51,200 or more per month) was removed from the chi-square test, the observed sample was similar to that of the expected sample, suggesting a bias towards wealthier people. In Nelson Mandela Bay, the reverse was true in that when the lowest income category (R1–400 per month) was excluded from the chi-square test, the sample was representative of the expected population, suggesting a bias towards poorer households.

Overall, 68.9% of respondents owned their property as opposed to renting, 83.3% lived in a house, 44.0% had a garden, 17.7% had a rainwater tank, 9.2% had a swimming pool, 6.0% had a greywater system, and 6.2% had a borehole.

The average water bill across all three municipalities was R460.61 per household per month and the median was R400.00. At the municipal scale, the average water bill ranged from R444.88 in the City of Cape Town to R482.48 in eThekwini (Table 3).

High confidence in respondents' water bills may be attributed to the fact that a large proportion of the sample pay their water bills directly to the municipality (82.7%), as opposed to it being part of their rent or building levies. Another plausible reason is that most respondents had a pay-as-you-go water meter (83.3%). Such a high degree of confidence in the estimated monthly water bills was beneficial to the study because these were used to calculate the initial bid for the double-bounded dichotomous choice questions.

The results from the generalised linear model for validating the double-bounded dichotomous choice questions are presented in three models in Table 4. The number of observations differed per model, since questions about education and income had ‘prefer not to say’ categories which were not included in the analysis. Model 1 represents the constant-only model, Model 2 includes several covariates, and Model 3 represents the estimation results when all non-significant predictors were removed from Model 2.

The results followed expectations in that the coefficients of the bid variable were negative across all three models, indicating that as the bid amount increased the probability of a respondent selecting ‘yes’ declined. After removing the non-significant covariates, Model 3 shows that the significance and magnitude of each variable did not change much compared with Model 2, verifying the robustness of the statistical model. Model 3 was considered the best model of the three because it had the lowest AIC value. Income, satisfaction with municipality, and household size significantly influenced WTP (Table 4). As expected, WTP was directly proportional to income, meaning that wealthier households were willing to pay more for water than poorer households. Similarly, those who were satisfied with their municipal water supply were willing to pay more for water compared to those who were dissatisfied or neutral. Interestingly, as the size of respondents' households increased, the probability of selecting ‘yes’ to a proposed bid increased. There was a significant difference between WTP in eThekwini compared to the WTP in Nelson Mandela Bay, but not between the City of Cape Town and Nelson Mandela Bay. On the other hand, education had no significant influence on WTP. It is, however, important to note that trust in municipality is probably correlated with satisfaction.

Based on the outputs of Model 3, the mean WTP per household per month in all three municipalities ranged from R392 to R513, and the unadjusted mean WTP was estimated to be R440 (Table 5). Median WTP for all municipalities ranged from R241 to R303, with an estimate of R272, while mean WTP was more than 2.5 times higher in eThekwini compared to the Nelson Mandela Bay (Table 4). Interestingly, despite mean WTP being far lower in the City of Cape Town compared to eThekwini, median WTP in Cape Town was higher.

To validate whether respondents were consistent in their responses, an open-ended maximum WTP question was asked. To improve data accuracy, these results were weighted according to the number of participants in each income category. The weighted average was highest for eThekwini (R464), followed by the City of Cape Town (R396) and Nelson Mandela Bay (R390) (Table 6). When the responses from all three municipalities were considered, the weighted average was R421. Although this value is slightly lower than the average WTP yielded by the double-bounded dichotomous choice model, it falls within the lower and upper bounds of the model and thus validates that respondents were consistent in their responses (Table 5).

For the purpose of this study, only households earning above R6,400 were of interest, as any household earning below this threshold qualify to receive indigent benefits from the government to help them pay for their service charges. Thus, for households earning more than R6,400, the range in mean WTP was between R493 and R691 in the City of Cape Town, between R344 and R450 in Nelson Mandela Bay, and between R499 and R667 in eThekwini. Depending on income level, this was 12–137% more than what households currently pay for water in the City of Cape Town and 32–73% more in eThekwini.

When compared to the corresponding water bill, WTP in eThekwini and the City of Cape Town was equal to 34% more than what residents currently pay for water in both municipalities. However, in Nelson Mandela Bay, households were not willing to pay any more for water than what they currently paying.

Aggregate WTP for existing water supply was R2.65 billion per year for the City of Cape Town and R1.58 billion per year for eThekwini (Supplementary Table S2). This translated to a consumer surplus of R779 million per year for the City of Cape Town and R250 million per year for eThekwini (Supplementary Table S3).

Although many countries have recognised the importance of restoring and protecting catchment areas to secure future water supply, there has been limited progress in implementation due to a lack of funding, particularly in developing countries. One financing instrument, which has not yet been mainstreamed, is that of water tariffs. This has been attributed to political obstacles that hinder the implementation and effectiveness of water reforms (Dinar 2000). For example, a political party may prevent water tariffs from increasing owing to fear of losing popularity among voters. However, set appropriately, the price of water can encourage users to manage their water consumption as well as generate revenues, which can fund water resource management and/or investment in ecological infrastructure. Since water is a basic need, reforming water pricing policies must be based on a comprehensive understanding of consumers' demand for water services and conservation.

Using three South African municipalities as a case study, this study set out to establish whether the additional revenue that could be generated from household WTP for water (i.e., consumer surplus) at the municipal scale is sufficient to cover the costs required to restore catchments supplying water to the municipalities in question. In two of the three municipalities, consumer surplus was more than sufficient to cover the estimated costs required to restore catchment areas supplying water to those municipalities. Factors that were found to significantly influence WTP included income, household size, and institutional satisfaction. This is a novel outcome in that it demonstrates to policymakers that there is significant potential to raise funds for catchment conservation through increasing water tariffs in high-income municipalities with good service delivery track records.

It is important to note that in this study, wealthier households were overrepresented relative to poorer households when compared with census data in both eThekwini and the City of Cape Town, while the reverse was true for Nelson Mandela Bay. As average WTP was potentially upwardly biased for eThekwini and the City of Cape Town, aggregate WTP was more reliably computed using a model that controlled for income category.

In accordance with previous contingent valuation studies in South Africa, this study reveals that across the three municipalities, household residents are willing to pay higher water tariffs for the quantities of water they are currently using, despite having high levels of dissatisfaction in water service provision (Kanyoka et al. 2009; Makaudze 2016; Nkoana et al. 2019). Of these, the WTP of households in this study correspond closest to that of Makaudze (2016) who measured WTP for water of residents living with HIV and AIDS in three districts (urban, peri-urban, and rural) and found it to be R429 per month (R570 at 2023 price levels). This amount is comparable to that of this study, where the estimate for WTP of households across the three urban municipalities was R440. However, given that the increasing block tariff system is employed in South Africa, the average WTP for water is not of great significance to the outcome of this study.

Of greater significance is how WTP varies across municipalities. This study found that households earning above R6400 per month (i.e., those not receiving any support from the government) were willing to pay 12–137% more than what they currently pay for water in the City of Cape Town, and 32–73% more in eThekwini, depending on income category. When weighted averages were applied, this was equal to 35% more than what residents currently pay for water in the City of Cape Town and 34% more in eThekwini. According to Cartwright (2021), financing ecological infrastructure interventions could be achieved through a 2.5% increase in the net price of water paid by industrial and domestic users in the City of Cape Town, and a 1% increase in eThekwini. Thus, this study shows that such increases would be more than tolerated by the non-indigent population. Interestingly, households in Nelson Mandela Bay were unwilling to pay any more for water than what they currently are paying, indicating that households are already paying their maximum, either because they simply cannot afford it or because they do not believe their money will be used effectively by the municipal government.

To understand why WTP differed across the three municipalities, it is necessary to consider the contextual findings of the study. At the time the surveys were conducted, the Nelson Mandela Bay was experiencing a water crisis. To encourage households to consume less water during the crisis, the municipality raised domestic water tariffs and imposed severe water restrictions on the population. This has had two noticeable impacts. First, relative to income levels, households in Nelson Mandela Bay are paying significantly more for water than the other two municipalities, which is evident when comparing the average water bill and income level of households in Nelson Mandela Bay to the other municipalities. By illustration, while the average water bill of households in Nelson Mandela Bay was similar to the other two municipalities (only ∼4% less than that of eThekwini and 4% more that of City of Cape Town), average reported income of respondents in Nelson Mandela Bay was 86 and 78% less than that of eThekwini and the City of Cape Town, respectively. Second, due to the perceived mismanagement of water supply, residents in Nelson Mandela Bay were extremely frustrated with the municipality. This was demonstrated through the high proportion of households in Nelson Mandela Bay who were dissatisfied with municipal service delivery compared to households in eThekwini and the City of Cape Town. Since residents in Nelson Mandela Bay were paying a lot of money towards a service they are generally dissatisfied with, they were not willing to entertain a price increase.

As expected, there was a significant positive relationship (p < 0.001) between WTP and income across all three municipalities. This provides validation of the survey design and is consistent with WTP studies generally (e.g., Halkos & Matsiori 2014; Makwinja et al. 2019; Van Song et al. 2019).

Results indicated that residents' level of satisfaction for the way the municipality manages water supply in their neighbourhood had a significant positive influence on WTP (p < 0.05). Thus, those who were more satisfied were willing to pay higher prices for water. This was anticipated since consumers tend to be willing to pay more for services they are satisfied with (Grafton et al. 2020). This is clearly the case in Nelson Mandela Bay, where levels of dissatisfaction were the highest. However, if the question was based on improved water supply instead of current water supply, the reverse would probably be true.

Another factor that had a direct relationship on WTP was household size (p < 0.05), which implies that households with more members were more likely to accept a given bid. A similar observation was made by Van Song et al. (2019). Although institutional trust also had a positive impact on WTP, the relationship was not significant. This finding is similar to that of Roldán et al. (2021) who found that developing countries are generally willing to pay more for water but do not trust their governments. In this study, nearly 59% of people did not trust that their municipality would use the funds appropriately if water tariffs were increased. Distrust is common in South Africa and is borne out of high levels of corruption in the country, which has resulted in frequent water interruptions, failing infrastructure, and poor water quality in all three municipalities (Makaudze 2016).

Increasing water tariffs to generate additional revenue for catchment conservation can only be implemented in areas where there is a positive consumer surplus. This study revealed that this was only the case for the City of Cape Town and eThekwini.

At present, water pricing cannot be used as a tool to secure funding for the restoration and maintenance of catchments supplying water to Nelson Mandela Bay, and alternative funding sources must be secured such as private–public partnerships, water funds, or through market-based instruments (e.g., Payments for Ecosystem Services schemes, biodiversity offsets, or carbon trading schemes) (Gómez-Baggethun & Muradian 2015). However, for the two municipalities that had a positive consumer surplus, there is potential to increase the water tariff to cover the costs of catchment conservation. It is proposed that between R325 and R375 million would be sufficient to generate water gains over the next 30 years in the City of Cape Town (TNC 2018; Webster et al. in prep). If the domestic water tariff was raised in accordance with WTP in this study, R779 million in additional revenue could be generated in one year, which is more than double of what is required to clear invasive alien plant (IAPs) species from the catchments that supply water to the City of Cape Town over 30 years. Similarly, the additional revenue that could be generated from untapped WTP in eThekwini (R250 million per year) would, over a 30-year period, be sufficient to clear IAPs from the Greater uMngeni catchment, which has been estimated to cost R250 million per year over five years by Cartwright (2021) and R1.23 billion over the next 30 years (at a discount rate of 8%) (Webster et al. in prep). While these estimates may be inflated, they do suggest that there is significant potential for raising tariffs and that people would be willing to pay quite a bit more than what they are currently paying.

To determine whether revenue generated from urban WTP can cover the costs of catchment management and conservation, the WTP of households was estimated using the contingent valuation approach. Surveys in three metropolitan municipalities of South Africa were conducted to elicit WTP at the local (municipal) scale. The results obtained from the double-bounded dichotomous choice model indicated that across the three municipalities, households had a higher WTP for water and that WTP was positively influenced by household income, age, and satisfaction with municipal water service delivery. However, when assessed separately, WTP differed significantly, largely due to differences in institutional satisfaction, household size, and income.

This study showed that there is significant potential to raise domestic water tariffs. However, the study has also shown that leveraging user fees hinges on functioning institutions. Satisfied residents who have some degree of trust in their municipality are more likely to be willing to pay higher prices for their water.

The authors would like to thank the Water Research Commission (WRC) for funding this project. The authors are grateful to Martine Visser from the School of Economics (University of Cape Town) for her comments on an earlier draft. The authors would also like to thank the two anonymous reviewers for their comments.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.