Are long-held beliefs and taboos on recycled water reducing in Southern Africa?¹

With increasing pressure on water resources in several parts of the Limpopo River Basin (LRB), the concept of beneficial use of treated wastewater has captured the attention of water resource managers (Petrie et al., 2014). As in several parts of the world, the concepts of reclamation, recycling and reuse are now recognized as key components of water resources and wastewater management in the LRB. To ensure the successful implementation of water recycling projects, it is imperative to align efforts to build new water treatment plants (or increase capacity of existing ones) with efforts to explore new end uses (Mainali et al., 2013) and further understand the potential end-users’ perception, willingness to use (WTU) and willingness to pay (WTP) for recycled water. After agriculture, the household sector is the second-largest water user in the LRB (Lombaard et al., 2015). Therefore, careful consideration should be given for conservation of more household water with recycled water.

The significance of recycled water, as part of water resources management, has long been acknowledged by several scholars (Angelakis & Bontoux, 2001; Po et al., 2003; Xiong et al., 2005). Owing to the advanced level of technology used in wastewater treatment, there is an increasing range of opportunities for putting recycled water to some economic use (Po et al., 2003; Menegaki et al., 2007). The benefits of using recycled water include protection of water resources, prevention of pollution, recovery of nutrients for agriculture, augmentation of river flow, savings in wastewater treatment, groundwater recharge, and sustainability of water resource management (Angelakis & Bontoux, 2001; Toze, 2006). Chen et al. (2014) argue that the use of recycled water for household laundry for instance can free a considerable amount of fresh water and reduce effluent discharge.

This paper aims to assess WTU and estimate households’ WTP for quality attributes of recycled water in the LRB. Valuation of attributes of recycled water is important especially for cases where water authorities and private companies producing recycled water are unable to overcome one or a set of attributes that will make recycled water at par with regular water. An example is colour of the recycled water – some recycled water plants are able to produce recycled water with at least the same level (or better) of scientific quality as regular water except for the colour of the water². In other cases it is the smell of the water. We apply a choice modelling approach to elicit households’ preferences for different quality of recycled water. The results of the study will benefit policymakers and private actors interested in investing in recycled water in the region. The next section of the paper provides information on water resources and demands in the LRB. We then review the literature on recycled water acceptability, present a conceptual framework and the research methodology, and then provide results, discussion and a conclusion.

The LRB is a complex transboundary system that supports 18 million people across four Southern African Development Community (SADC) Member States: Botswana, Mozambique, South Africa and Zimbabwe, and it has a total catchment area of approximately 408,000 km². It is an important agricultural area, has extraordinary mineral resources and is exceptionally rich from a biodiversity point of view (Petrie et al., 2014). The basin has approached water resource closure, and thus choices need to be made about the future management of the LRB system. Water scarcity in the basin has been well documented especially in the face of climate change and growing economic and population dependence (United Nations Environment Programme (UNEP), 2009; Petrie et al., 2014; Lombaard et al., 2015). Large urban centres such as Gaborone, Pretoria, Johannesburg and Bulawayo are the major users of domestic water within the basin. On average, the greatest water user sector in the four LRB riparian states is irrigation, which accounts for approximately 50% of the total water demand and up to 75% in some parts of the basin (Lombaard et al., 2015). The household sector is the second-largest water user in the LRB (Lombaard et al., 2015). The total estimated demand is presently about 65% of the total natural run-off generated from rainfall. Water allocation between upstream and downstream areas and among key water user sectors or domains is an important challenge for the future water management of the LRB. The LRB is one of the most vulnerable basins in the Southern African region due to its relative likelihood of being affected by resource shortages and climate-related risks, and also in terms of its limited capacity to adapt. The Southern African region's response to climate change over the last two decades has primarily focused on national policy changes and adaptation planning.

There is a general consensus in the literature that even though recycled water has provided sufficient flexibility to satisfy short-term freshwater needs, its uses are still quite limited. Uses of recycled water are mainly associated with non-potable purposes such as irrigation, industrial uses, toilet flushing and car washing (Mainali et al., 2013; Chen et al., 2014). However, there is empirical evidence supporting its use for potable purposes (Stoakley, 2013) or demand for goods that are produced using recycled water (Menegaki et al., 2007).

Several factors play a role in the acceptance of recycled water; the most important ones are: the stigma attached to it (disgust or ‘yuck’ factor), health concerns, the use for which recycled water is intended, the sources of recycled water (e.g. is it rainwater or toilet water?), trust of authorities, choice between recycled and fresh water, knowledge, attitudes towards the environment, environmental justice issues, socio-demographic factors and the price of recycled water (Po et al., 2003). Marks (2005) argues that the degree of public acceptance is also affected by the political context of a country.

As noted above, price is an important factor influencing the usage of recycled water. Price has been highlighted in the Australian National Water Quality Management Strategy (NWQMS) as an important determinant for community acceptance. The general expectation of households is to pay less for using recycled water as they perceive it to be of lower quality. Marks et al. (2003) found that even though respondents valued recycled water and acknowledged the benefit of water conservation, they were mainly interested in the personal cost benefit involved in using this cheaper alternative to potable water. Stoakley (2013) also found a uniform concern, among respondents, about the price of recycled water. Specifically, the price reflects how households’ trade off their perception and beliefs on recycled water for a reliable source and a lower water bill.

Mainali et al. (2013) argue that ‘successful implementation of a wastewater reuse project depends not only on its technical and environmental feasibility, but primarily on the support and the acceptance from the general public’. Lazarova et al. (2012) claim that ‘In addition to economic, social and environmental benefits, a distinct benefit of water reuse is the steadiness of water supply for both household and local industries, which is superior to rainfall-dependent water sources.’ Combining recycled water with other water resources could potentially reduce the ecological footprint of water sewerage and drainage systems (Anderson, 2003).

For i = 1, 2 … N; j = 1, 2 … J and t = 1, 2 … T. X_ij is a vector of observed attributes for choice j and P_ij is the price of alternative j that individual i faces.

Households in the LRB across three riparian states were surveyed. Specifically, in Mozambique we surveyed households in the Pafuri-triangle, in Zimbabwe households in the Upper Umzingwane, and Botswana households in the Upper Limpopo area of the basin. Selection of respondents in all three locations was random. Households were approached by trained enumerators in their houses, with consent to participate in the survey given before the households were interviewed. The purpose of the study was also explained to the households. All the households interviewed were provided with the basic facts and definition of recycled water.

Households were given a list of activities (potential uses of recycled water) and asked whether they would be willing to use recycled water for them. These include domestic use for food, bathing, washing clothes, washing car, filling pool, flushing toilet, watering garden and other. The description of quality of recycled water and the definition were the same across different uses.

The questionnaire is divided into six sections. Section One asks for basic household information to identify the person who is interviewed – it elicits information such as gender, age, education level and occupation. Section Two elicits the current and expected water demand and availability situation. This section asks questions on major water sources and uses by households; highlighting where households get their water; asking about water demand sources such as borehole, river, rain and tap water. Several Likert-scale questions are asked on the quality of water supply. Section Three of the questionnaire elicits attitude towards efficient use of water including investments such as retrofitting to reduce the use of water by the households. The section also elicits the preference for recycled water usage at the household level asking about WTU for different domestic usages such as drinking, bathing, etc. Section Four elicits agriculture usage of water in the basin, asking about irrigation including surface water and groundwater and the source of the water for it. Similar questions are also asked for fisheries and livestock production. Section Five gathers household-level information including socio-economic status, environmental affiliations and financial information. The last section of the survey presents the choice modelling questions. This section presents three choices for households to rank based on the different combination of attributes of recycled water. Attributes of recycled water presented include taste, colour, odour, salt level, suspended solids and relative price of the recycled water. Table 1 presents control variables, and their descriptions, used in the econometric model. Table 2 presents the quality attributes, of recycled water, used in the WTP econometric model.

The proportion of households willing to use recycled water for food is 36.45% and a larger proportion (70.26%) are willing to use recycled water for flushing toilet (Table 3).

We tested the role of different socio-economic characteristics on the WTU recycled water. For example, what is the role of education in changing the beliefs of households on using recycled water? The results from the survey show that about 24% of households where the household head has no formal education are willing to use recycled water (Table 4). This is lower than the percentage for primary school education holders (31%), or high school holders (30%). There is some indication from the survey that education increases the WTU recycled water of households in the LRB. We also looked at the differences in WTU recycled water by occupation. Interestingly, the percentage is lower among farmers than other types of workers such as technical professional staff and civil servants.

The WTU recycled water by the source of water for the household is also evaluated. This is an interesting analysis that can indicate the potential for trade-off by households. With a higher quality of water available to a household, one will expect that the WTU recycled water will reduce. The results show that households with a borehole with a machine are least likely to use recycled water – they have already invested in a water supply mechanism and trade-off with recycled water is unlikely. However, we see a 98% WTU recycled water by households in the LRB whose main source of water is the river (Table 5). Households using a river as their main source account for 4% of the sample.

We also compared WTU recycled water and ranking of quality of the main source of water. We found that households that do not consider their main source of water safe for drinking have a higher likelihood of using recycled water (85%) compared to those who perceive their main source as safe for drinking, of which 72% were willing to use recycled water (Table 6). Another comparison of perception of water quality of the main source (colour, taste, and smell of the water) with WTU recycled water also indicates similar results. A total of 91% of households that strongly agree that recycled water should be used for potable uses are willing to use the recycled water, while only 73% of households that disagree or strongly disagree that it should be used for potable uses are willing to use it. Table 7 presents households’ perception towards potability of water from the main source and WTU recycled water by location. In Pafuri-triangle, 44.35% of the households who strongly agree that their main water source is safe for drinking are willing to use recycled water.

We also asked a question on the perception of water supply in the LRB and WTU recycled water (Table 8). The results show that households that perceive that water supply will be lower in the future are more likely to use recycled water. Lastly, households that indicated that they would not be able to use the same amount of water if the price of water were to double or triple are more likely to switch to recycled water (85%) than those who can afford an increase in prices (Table 9). We present additional details on the relationship between different perceptions of water efficiency and WTU recycled water by households surveyed in the LRB.

Households that agree that water is a scarce resource also reported to be willing to use recycled water (Table 10). The same argument holds for availability of water in the future and responsibility of households to use water efficiently (Table 11).

Table 12 contains output from the multivariate probit model for three sets of potential uses of recycled water. For ease of interpretation, individual categories of education level and occupation were each represented by a dummy variable (D = 1 if the respondent answered in the affirmative for that particular category, zero otherwise). Table 12 presents three models; Model 1 estimates WTU recycled water for food, Model 2 estimates WTU recycled water for bathing, washing clothes or filling pool, finally, Model 3 estimates WTU recycled water for washing car, watering the garden or other uses. We report results for these three groups, however we also provide results, in the Appendix Table A1 (available with the online version of this paper), without grouping the data. The results indicate that there is gender disparity in the WTU recycled water for food (Model 1) and bathing, washing clothes or filling pool (Model 2). The probability that female respondents were willing to use recycled water for food was 0.171 higher than that of their male counterparts; however, female respondents were less likely to be willing to use recycled water for bathing, washing clothes and filling pool (probability was 0.145 lower than that of male respondents). Age and role of respondents in the household are associated with WTU recycled water for bathing, washing clothes or filling pool. A one-year increase in age decreases the probability that the respondent would be willing to use recycled water by 0.011. Respondents who were also household heads were 25.9% more likely to be willing to use recycled water than those who were not heads of their households. For education level, the tertiary category is the benchmark category; therefore the reported marginal effects for the rest of the categories are relative to the category of households who attained tertiary education. The results show that respondents who had no formal education were found to be less likely to be willing to use recycled water for washing car, flushing toilet, watering garden or other uses compared to those with tertiary education. Households who had high school as their highest qualification were also found to be less likely to be willing to use recycled water for the above-mentioned uses; however, they were more likely to be willing to use recycled water for food, and for bathing, washing clothes or filling pool. With regard to occupation, the benchmark category is ‘technical/professional/managerial-civil service’. The estimates of the correlation between the equation error terms show correlation between the WTU recycled water for food and for bathing, washing clothes and filling pool but no correlation in the errors between WTU for food and for washing car, flushing toilet, watering garden and other uses. There is correlation between the errors in the bathing, washing clothes and filling pool equation and using the recycled water for washing car. The unobservables that influence the WTU recycled water for food also drive the use of recycled water for bathing and washing. This result indicates that there are other factors not captured in the model that also significantly jointly drive the WTU recycled water for food and bathing but not for food and washing car.

In this section we present results from the choice experiment that estimates consumer preferences for different attributes of recycled water. We present results from a RUM framework that estimates preferences of households for recycled water and the attributes of the water that they care about with estimated WTP for those attributes.

Households were presented with three options in a choice experiment to understand preferences for different attributes of recycled water. Figure 1 shows a sample question representing a choice occasion presented to households. The value of these attributes is important for pricing of recycled water, especially in a developing country where water authorities and government may be struggling with some aspects of the technology of recycled water. For example, discussions with the water board in Botswana highlight that in some cases recycled water production may have the same water quality as other types of water but they just could not get the colour of the water right. Therefore, while nothing is wrong with the recycled water scientifically, the colour attribute of the water is different from that of regular water. This has led to the water authority not being able to supply the recycled water.

Following the RUM methodology as presented in Abidoye et al. (2011), we estimate household WTP for attributes of interest in recycled water, which is useful as the technology and marketing of recycled water change. We apply the random parameters logit model that is a generalization of the standard logit model over the parameter distribution. This model is preferred because it takes into consideration the correlation between the alternatives.

Using the data collected in the basin, we estimate the demand for recycled water in the LRB. Before explaining the results, we explain the model and the results table. To compute WTP for each attribute, the coefficient of the attribute of interest is divided by the negative of the coefficient of relative price. Table 13 shows that households are willing to pay for improvement in colour and taste of recycled water. On the contrary, they are willing to accept a compensation for odour, high suspended solids, medium suspended solids, high salt and medium salt.

Table 14 presents estimates for the mean and standard deviation of the distribution from the mixed logit model. The estimated parameters can be interpreted as a vector of coefficients of the observed attributes that vary over individuals in the population. The estimated parameter captures heterogeneity in preferences for the attributes that can be characterized by the mean and standard deviation. The first column presents the mean estimates and the second column is the standard deviation of the distribution³.

The first result (relative price) is estimated as a constant because it represents the marginal utility of income.

Households’ perception of colour of water from the main source is one of the significant factors influencing their WTU recycled water. Households who are happy with the colour of water from the main source are less likely to be willing to use recycled water for any of the presented potential uses. This result is also reinforced by the high WTP for improvement in colour of recycled water. Male respondents were found to be less likely to be willing to use recycled water for food, however more likely to be willing to use recycled water for flushing the toilet, watering garden or other uses compared to their female counterparts. Respondents who were also heads of the household were found to be more likely to be willing to use recycled water compared to those who were not heads of the household. An increase in average income reduces the likelihood that households would be willing to use recycled water for washing clothes, washing car and watering garden. Age reduces the likelihood of WTU recycled water for bathing, washing clothes or filling pool. This result is consistent with the prediction of McKay & Hurlimann (2003) that the greatest resistance towards water reuse would be from people aged 50 years and above.

The results show that price is a significant component of the demand for recycled water. The negative coefficient of relative price (Table 13) indicates that the higher the price of the recycled water, the lower the likelihood of using recycled water by households in the basin. Colour is an important attribute of water that households care about. The less colouring the recycled water has, the higher the likelihood of using the recycled water. With the standard deviation, not significant, the results show that households uniformly agree that colour is an important component of the demand for recycled water in the basin. This result is in consonance with the framework for the assessment of new end uses in recycled water schemes developed by Chen et al. (2014), which postulates that, among other factors, public perception of colour is a threat to the feasibility of water reuse projects. Taste is also significant in determining the demand for recycled water although the evidence is not strong on this with the parameter only significant at the 10% level. Odour is not an important determinant of recycled water usage in the region, as the results show. Total suspended solids are particles that are larger than 2 microns found in the water column. The results show that households care about this, with usage of recycled water reducing as suspended solids in the water increase. Most importantly, households seem to be able to live with medium suspended solids, with the standard deviation of 0.36 indicating that there is heterogeneity on how households perceive medium suspended solids, but unanimously see high suspended solids' presence as a problem. Lastly, we look at salt. High and medium salt are also considered bad and reduce the demand for recycled water. The higher the salinity in the recycled water, the lower the likelihood of using the recycled water.

We also looked at how much households are willing to pay for these attributes to be able to estimate WTP for recycled water. The results show that households are willing to pay less for recycled water in general but are willing to pay for improved quality up to 76% of the price of standard water for recycled water with no colour. The WTP for recycled water with no taste is 42% of the price of standard water. Water quality issues such as high suspended solids and high salt are attributes that households find non-negotiable in their preference for water and are considered as a ‘bad’ good. Households are willing to accept compensation up to 160% of the cost of standard water for high suspended solids and 131% of the standard cost in the case of high salt concentration.

With increasing pressure on the current water supply mechanism and infrastructure in the LRB due to factors such as climate change, a combination of demand-side and supply-side interventions is essential to cope with the impact on households. This paper supports appraising adaptation option effort by policymakers in the region to reduce water resource vulnerability. One potential intervention that is gaining prominence is the use of recycled water. This study identifies the factors that affect the demand for recycled water by households in the LRB and estimates WTP by households for different attributes of recycled water. The results are important for measuring the potential for trade-off between long-held beliefs about recycled water and current water vulnerability. This is also useful for policymakers in the adoption of policies that can nudge households to increase adoption of recycled water as an adaptation strategy. Specifically, recycled water has some preconceived benefit – a priori negative – which can limit adoption. The type of nudge that may be useful in the region is predicted by this model by eliciting the preferences of households in the region and the attributes of recycled water that are important for adoption and WTP. This study explores these attributes and provides evidence-based recommendations to policymakers.

Key conclusions and recommendations that emerge from the analysis include the following:

• 1.

  Recycled water is valued lower than water from the main source by households in the LRB. Utility decreases by about 1.8 utils for every 10% relative difference in price between the two water sources. This is important for pricing of recycled water in the countries – subsidies may be useful to nudge households to use recycled water in the region.

• 2.

  Depending on what household use of recycled water policymakers may want to target, the factors that influence the demand vary. The results in this study show that the following factors presented in Table 15 are important for the demand for recycled water in the LRB.

• 3.

  To prioritize investment in recycled water technology in the basin, this study shows that the following attributes are important for household demand for recycled water in the LRB:

  • a.

    Colour, price, suspended solids and salinity level of recycled water are the major attributes that households care about in the basin. Recycled water that can get these attributes right will be acceptable to households.

  • b.

    Colour is an important attribute of water that households care about. The less colouring the recycled water has, the higher the likelihood of using the recycled water. There are cases where recycled water is scientifically not different from water from the household's main source except for differences in the colour of the water⁴. Policymakers should pay attention to this and invest in technology that can improve the colour or interventions that can help households overcome the colour bias.

  • c.

    The evidence on socio-economic characteristics associated with WTU recycled water is useful in designing recycled water use campaigns. It gives an indication as to which group of households should be educated on recycled water. Following the results of this study, it is recommended that campaigns should target household heads. Households involved in businesses such as transport are willing to use recycled water. This presents an opportunity for public–private partnership in the distribution of recycled water to these businesses.

• 4.

  In general, households are more willing to use recycled water for non-potable uses such as washing car, flushing toilet, watering garden or other uses. This is a step in the right direction with the potential for reducing the pressure on current water usage if households can rely on standard water sources for drinking and household consumption only. This by itself may require investments that may include the need for two separate pipelines with one dedicated to recycled water. The logistics and modality for having two different water sources may outweigh the benefit of it. However, recycled water can also work if awareness campaigns and exchange between countries is encouraged (for instance, learning from Namibia on how recycled water has been streamlined into major water usage patterns in the country). Mainstreaming the topic of recycled water into the school curriculum can also help with the awareness. Institutional capacity development may also be needed as a form of education to improve awareness and to develop improved technology.

We acknowledge the support of USAID Southern Africa, LIMCOM Secretariat, households and farmers met during the survey, national institutions and partners in riparian countries that contributed to the study. The data collection was carried out by Global Water Partnership and Center for Environmental Economics and Policy in Africa (CEEPA). We thank Armand K. Houanye, Andrew Takawira, and Michael Ramaano for their support in the data collection. Research fund support is from the National Research Foundation of South Africa.