Identification of pathways that lead to continuous or intermittent water supply by conducting a qualitative comparative analysis of rural water utilities in China

Households having access to an adequate and clean water supply is closely linked to most of the components of rural sustainable development, including poverty reduction (Saleth et al. 2003), social fairness (Tsitsifli et al. 2017), and human health (Martin et al. 2007). Ensuring ‘available when needed’ water supply (i.e. 12 h) for household use is helpful to the goals of the Millennium Development Goals (MDGs) (i.e. target 7c) (WHO/UNICEF 2015) and the Sustainable Development Goal (SDG) targets (i.e. SDG Indicator 6.1.1) (UN 2015). In the last decade, the drinking water supply has been greatly improved due to the intensive efforts of governments and non-governmental organisations. By the end of 2020, 74% of the global population includes 60% in rural and 86% in urban, had used safely managed drinking water services (i.e. piped drinking water on premises, available when needed and free of contamination) compared with the 70% in 2015 (WHO/UNICEF 2021).

Intermittent water supply (IWS) (i.e. non-24 h supply) and continuous water supply (CWS) (24 h supply) are two patterns of piped water supply. In IWS systems, water supply is unavailable for periods of time, and access to water ranges from predictable to unreliable (Galaitsi et al. 2016). Compared with CWS systems, IWS systems are intensively criticised for their insufficient water, inconvenient supply times, and additional costs for pipeline repair due to the lack of water inside (McIntosh 2003). IWS also causes hygiene risks of waterborne diseases due to water storage (Mintz et al. 1995), insufficient water for sanitation services, and influenced water usage behaviour for hygiene purposes (Fan et al. 2014) Despite its disadvantages, IWS is used because of low investments, easy management and effectiveness of controlling water demand (Faure & Pandit 2010). This system remains prevalent in less developed countries and serves at least 300 million people worldwide (Kumpel & Nelson 2016).

The reasons for adopting IWS vary across countries. Katuwal & Bohara (2011) found a poorly maintained distribution system and inefficient service as the main reasons for IWS in Kathmandu, Nepal. Stoler et al. (2012) verified water leaks in supply systems as the reason for IWS in Accra of Ghana. Souza et al. (2022) revealed that serious water shortage due to low rainfall prompted São Paulo in Brazil to adopt IWS from 2014 to 2016. Simukonda et al. (2018) identified poor governance, poor system management and operation, demographic and economic dynamics and unplanned system extensions, and lack of customer awareness as the reasons for IWS in Lusaka City of Zambia. The causal factors of IWS in different regions vary even in the same country. Andey & Kelkar (2007) found that the broadest network distribution of water supply in Ghaziabad and Panaji of India cannot meet the demand for CWS. Dutta & Tiwari (2005) confirmed low prices, wasted use, and insufficient fund support of water utility as the main reasons for IWS in Delhi.

Based on the global survey by the Intermittent Water Supply Specialist Group of the International Water Association, Farmani et al. (2021) identified intermittent electricity, infrastructure capacity, and water resource availability as the main causes of IWS in most water distribution systems. Totsuka et al. (2004) presented three types of problems (i.e. poor technical management, economic scarcity, and absolute scarcity of water resources) that can cause IWS. Simukonda et al. (2018) reviewed literature relevant to IWS and its causal factors and found that the interplay among political, social, economic, natural, and technical factors contribute to IWS. The factors causing IWS can be grouped into capacity (i.e. insufficient water source, facilities, and supply), economic (i.e. low profits and investments and high costs), and policy (i.e. lacks subsidy, insufficient policies, or misjudgements towards IWS) constraints. Each IWS system has several causal factors which interact and form configurations. Farmani et al. (2021) and Simukonda et al. (2018) highlighted these interactions among causal factors and call for interdisciplinary approaches to resolve these problems. Therefore, the causal factors and particular configurations of IWS and CWS must be identified for public policy making.

In China, the billion-dollar investment of the Mother Cisterns Project (2001–2011) and the Chinese Safe Drinking Water Project (2005–2015) greatly improved the water supply and enabled its broad access (Fan et al. 2013). Thousands of small-scale water utilities are established and scattered in rural areas. The coverage of piped water on premises in rural China was around 82% in 2020 (Li et al. 2021). Some water utilities operated by town and country governments usually serve large populations and are well-supported by financial resources and policies (Jiang et al. 2022). By contrast, those water utilities operated by villager councils and private individuals are often fragile, lack financial support, and usually supply one to three villages (Fan et al. 2014).

Similar to other developing countries in South Asia, India, and Latin America, many small-scale water utilities in China deliver water only in the mornings and evenings (IWS) for various reasons, including a lack of water availability and supply restrictions (Davies & Westgate 2018); to reduce expenditures (Li et al. 2020), tariff structure and pricing and consumer behaviour (Fan et al. 2013; Fan et al. 2014). Galaitsi et al. (2016) identified 47 conditions (factors) that may cause IWS systems. For each village-owned water utility, the causal factors can be similar or different depending on the local background and its characteristics. These causal factors are interconnected into complex causal networks and form several combinations (pathways or configurations) that can cause IWS. However, the causal pathways (configurations) of water utilities adopting IWS and CWS systems in rural China have not been sufficiently examined.

A survey of village-level water utilities in rural areas of Shaanxi Plain, China was conducted to (1) determine the IWS and CWS systems in rural China, (2) identify the causal factors and configurations (pathways) of water utility adopting IWS and CWS systems, and (3) propose appropriate solutions and policies for reducing IWS and improving CWS in rural China and provide a reference for other developing countries.

fsQCA was developed by Ragin (2000) and is an analytical technique used to identify logical connections (relations) between combinations of causal conditions (conjunctural causation) and an outcome from a limited number of data and cases. Compared with the conventional correlation and regression-based method (CRM), fsQCA has merits and is useful in exploring intertwined relationships among a set of multiple variables (Fiss 2011).

In CRM, the relations between independent and dependent (outcome) variables are always linear and symmetric (Skarmeas et al. 2014). In reality, most observed relationships are often asymmetric and non-linear. fsQCA analyses a set of relationships among causes to find the causal conditions (pathways) that lead to a given outcome. This method is used as a supplement to CRM and captures three aspects: asymmetry, equifinality, and causal complexity of variables (Skarmeas et al. 2014).

fsQCA is widely applied in sociology, including project and policy analysis (Kirchherr et al. 2016), environmental science (Bothner et al. 2019), business research, tourism, and hospitality management (Yadav et al. 2019), institutional analysis and organisational behaviour (Meyer et al. 2018). Despite a few representative publications on water quality monitoring (Peletz et al. 2018), water supply delivery (Gasparro & Walters 2017), and sustainable water supply and climate change (Pahl-Wostl & Knieper 2014; Kaminsky & Jordan 2017; Marks et al. 2018), this technique has rarely been employed in water resource management.

fsQCA analyzes a set of relationships to identify all the necessary and sufficient conditions that lead to a specific outcome (Ragin 2000). Necessary conditions must be present when the outcome occurs, but their presence does not guarantee the outcome occurrence. Sufficient conditions are indicative of the outcome but are not the sole conditions leading to outcome occurrence; other sufficient conditions may coexist for the same outcome. For example, the ‘water conservation behaviour’ of an individual can be a sufficient condition of ‘water consumption reduction’ but is not the only condition leading to ‘water consumption reduction’. Popularising water-saving applications can also lead to ‘water consumption reduction’.

The Shaanxi Plain is a well-developed agricultural zone that has a large number of villages and small-scale water utilities. We first identified rural water utilities from the database of the Safe Drinking Water Project and Planning (2008–2012) of Shaanxi Province and then selected three sites from west to east (i.e. A, located in Baoji District; B, located in Yanglin–Xianyang District; and C located in Weinan District) with similar terrain, climate and socioeconomic features as sample sites (Figure 2). A total of 3,157 water utilities located in the Shaanxi Plain are owned by the county, town, or village governments. The water utilities which are owned by a county of town government are often located close to a city or town with a large population and are often well-funded and managed (Jiang et al. 2022). By contrast, utilities owned by a village (i.e. village council) are often located in relatively remote rural areas with scattered supplies serving a small population and have poor funding and management (Whittington et al. 2009). The latter is fragile. Two types of operators exist in village-owned utilities: village councils and private individuals (i.e. private contractors). Given the goal of providing sustainable water supply to low-income families and remote rural areas, our survey focused on the water utilities owned by villages.

The supply mode of water utilities in rural areas is influenced by many factors, including water resource availability and water use efficiency (Cubillo 2004), water supply management and facility support (Marks et al. 2013), economic support (Gasparro & Walters 2017), and government and community participation (Walters & Javernick-Will 2015; Simukonda et al. 2018). Here, water resource availability was not considered because of the sufficiency and stability of groundwater as a resource. Instead, we used water-saving initiatives for the efficient use of water resources because of the limited supply capacity from pipelines. Eight distinct potential causal conditions for the water supply mode are identified based on existing literature. (1) Contracted water utility indicates the operator of a water utility (i.e. by village council if not contracted or by private individuals if contracted). Different operators usually adopt varying strategies for water supply management because of their different roles and responsibilities (Renzetti & Dupont 2003; Bel 2020). (2) Water-saving initiatives measure whether a water supply system employs some water-saving programmes or incentives to encourage water conservation among villagers, and are correlated with good outcomes of desirable water supply like in Hong Kong, China (Yue & Tang 2011) and Abuja, Nigeria (Abubakar 2018). (3) Government subsidies measure whether a water utility receives subsidies from the government or community and are linked to reinforced financial support for water supply, such as in Kenya (Acey et al. 2019) and Mexico (Morales-Novelo et al. 2018). (4) Population served by supply measures the total number of people served by a water utility and is correlated with the costs of managing the utility. (5) Annual household income measures the degree of affluence in a family and is highly related to water consumption, water price formulation, supply service (i.e. hours of water supply), and management of the water project like in the 25 largest cities in the United States (Teodoro 2018) and Duhok, Iraq (Hussien et al. 2016). (6) Meter-based billing refers to whether water consumption is measured quantitatively (with a water meter) or qualitatively (without a water meter) and is correlated with waste water and inequity consumption and eventual harm to sustainable water supply, such as in Yangon, Japan (Soe et al. 2020) and Turkey (Selek et al. 2018); (7) Villager participation measures how much villagers are involved in the planning, implementation, and management of the water utility and is correlated with good outcomes of water supply projects, such as in Karnataka and Uttar Pradesh in India (Prokopy 2005). (8) Convenient transportation measures the distance between a village and a town. A village close to town possibly has a large population, and high income and might receive attention from the government (Zheng 2018).

Causal conditions and outcomes of the water supply mode of data collection in the survey are presented in Table 2. The outcomes of supply mode are (1) CWS, that is, water supply for 24 h and (2) IWS, that is, water supply for less than 24 h/day. Supplementary Material, Table S1 provides a correlation of eight causal conditions. There exists collinearity among the conditions, it is suitable for fsQCA to explore intertwined relationships among them.

Table 2 shows the descriptions of the eight causal conditions and outcomes of water supply mode. Given that the fuzzy set should range from 0 (full non-membership) to 1 (full membership), some data should be transformed and calibrated into values of 0.0–1.0 to correspond with external standards. The contracted water utility, government subsidies, villager participation, meter-based billing, and water-saving initiatives were collected as binary factors and did not require additional transformation. For population served by supply, annual household income, and transportation convenience, we set the value of 1 for full membership (i.e. fuzzy score = 1), 0 for full non-membership (i.e. fuzzy score = 0), and 0.33 and 0.67 for partial membership (i.e. fuzzy score = 0.33, 0.67). The criteria of thresholds for fuzzy scores were based on the data distributions (Table 2). With the above benchmarks, the original ratio or interval-scale values were transformed into fuzzy membership scores (Table 3).

All data were coded and analysed on the fsQCA software (version 3.0) developed by Ragin (University of California) to explore the conditions and combinations of conditions that are necessary or sufficient for IWS and CWS systems. Two ‘goodness-of-fit’ parameters, namely, the value of consistency and coverage, were measured to identify the necessary and sufficient conditions. Consistency (analogous to correlation) measures the extent to which a causal condition (or causal combination) is observed when the outcome occurs. Coverage (a measure of effect size) measures the probability that a condition (or a combination of conditions) is absent when the outcome does not occur. QCA provides accurate information on how causal configurations explain a certain outcome by using consistency and coverage (Woodside 2010). Considering that coverage reflects how much the outcome is explained by each causal combination (solution) as a whole (Ragin 2008), we did not set a specific threshold for coverage for a causal combination. According to Schneider et al. (2010), Schneider & Wagemann (2010), and Stroe et al. (2018), we set the thresholds of consistency as 0.90 for necessary conditions and 0.75 for sufficient conditions.

Over half of water utilities (20) are contracted by private individuals, and the rest (18) are operated by local village councils. Most utilities are self-sufficient, and some (15) receive subsidies from the local government and additional financial support from villagers (14). The majority of water utilities lack policies to encourage water conservation. All the water utilities (38) installed water meters and 44.7% bill by using water meters, and the rest bill using a fixed amount or by the population of a household (Table 3).

The necessary conditions of each causal condition on the outcome (i.e. IWS or CWS) were analysed through fsQCA. Table 4 presents the consistency scores of individual causal conditions for IWS and CWS ranging from 0.167 to 0.833 and their coverage scores from 0.200 to 0.800. No condition exceeded the consistency threshold of 0.90 (Schneider & Wagemann 2012), indicating that no individual condition alone leads to IWS or CWS (see Supplementary Material SI1).

We performed a configuration analysis with parsimonious solutions by using the truth table to determine the possible conditions of the CWS outcome. Three causal configurations that lead to CWS were obtained (Table 5 and Supplementary Material SI2). In fsQCA analysis, coverage represents the degree to which a causal configuration accounts for the outcome and reflects how much of the outcome is explained by a solution pathway as a whole (Hsiao et al. 2016). The overall coverage is 0.768, indicating that the three combinations account for approximately 76.8% of the membership of CWS. The remaining 23.2% of variances (other solution combinations) leading to CWS remain undiscovered. The three configurations have consistency scores of 1.00, which is higher than the threshold value of 0.75 recommended by Ragin (2000) for sufficiency.

For easy configuration description, we use the symbol ‘*’ that signifies ‘AND’ and ‘∼’ that represents the absence (i.e. low level) of a given condition (see Supplementary Material SI2). Configuration 1 (government subsidy * meter-based bill) shows that the water utilities can achieve CWS through high government subsidy plus meter-based billing. Configuration 1 interprets 13.4% variances that lead to CWS. Configuration 2 (government subsidy * villager participation * ∼population served by supply) indicates that for water utilities with small population coverage, government subsidy and villager participation in water supply management can lead to CWS. Configuration 2 interprets 30.1% variances of CWS. Configuration 3 (∼contracted water utility * annual household income * water-saving initiative) indicates that for rich villages (i.e. have a high level of annual household income), village operated water utility and water-saving programmes can lead to CWS. Configuration 3 interprets 25.1% variances of CWS.

Four configurations that lead to IWS were obtained (see Supplementary Material SI3). The overall solution consistency is 0.888, and the overall solution coverage is 0.871. The remaining 12.9% variances leading to IWS remain undiscovered. Configuration 1b has a consistency score of 0.689, which is below the recommended threshold value (i.e. 0.75) for sufficiency, and therefore was excluded (Table 6). The remaining three configurations are as follows. Configuration 1a (∼ meter-based bill * ∼villager participation) interprets 20.3% variances that lead to the IWS mode. Configuration 1a shows that an unmetered bill system (i.e. water bill charges based on family members or fixed bill, not by records of water meters) and lacking villager participation in management will lead to the IWS mode. Configuration 2 (contracted water utility * ∼government subsidy * ∼population served by supply) indicates that for water utilities contracted by private individuals and covering a small population, IWS will occur when government subsidies are lacking. Configuration 2 interprets 20.4% variances of IWS mode. Configuration 3 (∼government subsidy * ∼annual household income * ∼water-saving initiative * ∼distance from the market) indicates that IWS occurs in remote and poor villages without subsidies and water-saving participation. Configuration 3 interprets 7.4% variances of IWS mode.

This study examines how the affecting factors synergistically influence the water supply mode. We evaluate the six configurations (CWS: three configurations and IWS: three configurations) of eight factors affecting the water supply mode in rural China. Villager participation, water-saving programmes, government subsidies, and meter-based billing are highlighted as crucial for improving the water supply mode.

Villager participation is important for water utilities to obtain CWS mode. When operating water utilities, water managers have to deal with the multiple requirements of different groups of villagers to implement water management strategies. The different roles of villagers and water managers often impair policy making and implementation. Our finding aligns with Jorgensen et al. (2009) that consumer participation is beneficial for policy making and implementation through information sharing between consumers and water managers. Meanwhile, Mansaray et al. (2017) identified that respondents usually do not trust and follow the policies of service providers because of a lack of participation, especially those in the public sector due to corruption, incompetence, and dishonesty. Hartley (2006) confirmed that public participation enables information sharing between the public and water supply sector and should be considered in current and future water resource management. For utilities that serve a small population and receive subsidies, using those subsidies wisely is important for achieving CWS mode. Villager participation is essential for realising efficient strategies for using the subsidies wisely.

For village operated water utilities, the water-saving-based path could lead to the CWS mode. Although the supply capacities of water utilities are designed to meet the demand of villagers in rural China, water supply limitations may result in unsatisfied demands. (Li 2003). Therefore, water-saving programmes should be highlighted in urban and rural areas. The water use in rural areas is more complex than that in urban ones because of large outdoor water consumption, such as for livestock and homestead vegetables. Fan et al. (2013) found that outdoor water consumption accounts for over half of the total water consumption of rural households. Compared with indoor water (i.e. for kitchen, laundry, and personal hygiene), outdoor water is more freely and arbitrarily consumed. Thus, conservation targets should include outdoor water consumption, even for villages under IWS mode.

Implementing water saving measures can be difficult, particularly in rural areas. Our concern aligns with Fan et al. (2013) that traditional culture, habits, lack of information, and lack of trust by residents can hinder water conservation practices. Jorgensen et al. (2009) found that trust among neighbours and institutions enhances the willingness of people to save water. Liu et al. (2016) showed that information transparency regarding water use and management is important for water conservation through forming institutional trust and knowing the amount of water consumed and bills. In this study, we also found that improving institutional trust through villager participation and information transparency can support water conservation.

Economic-based path including government subsidy and meter-based billing is the main factor for water utilities adopting CWS. Rural water supply systems are particularly reliant on subsidies given the low density of villagers, lower per capita withdrawals, and often lower tariffs (Whittington et al. 2009). In rural China, the cost of a piped water supply system averages 30 USD per person only for installation (Shuchen et al. 2004). Apart from costs for installation, additional costs for facility maintenance and water provision (i.e. staff, electricity, and facilities depreciation costs) are also needed to operate a water utility. However, the water bill is around 6 USD per capita per year with water price being approximately 0.20–0.29 USD/m³ (Fan et al. 2014) and water consumption per capita per day is usually less than 70 L (Keshavarzi et al. 2006; Fan et al. 2013). Water bills from villagers are usually insufficient for operations. Therefore, investments and subsidies from the government, communities, and villagers are crucial. For low household income and water bills, government subsidies are the sole solutions for sustainable rural supply.

An effective billing system is important for the efficient operation of water utilities. However, in rural areas, water supply systems often adopt unmetered billing. This study found that meter-based billing is beneficial for CWS. Without a meter-based bill, residents need not pay for additional water used. The ‘tragedy of the commons’ in water use is unavoidable. Our finding aligns with Kumpel et al. (2017) which found that multiple intentional and unintentional wasted uses can be identified from various household activities in an unmetered supply system. Gan (2000) also found that in Calgary and in Edmonton, Canada, an unmetered household uses approximately 50% more water than a metered household. In Ontario, users with individual meters consume approximately 15–27% less water than those with common meters (Koop et al. 2015). These findings indicate a large amount of water wastage in an unmetered bill system. On the contrary, meter-based billing allows users to know how much they consume and how much they should pay. This approach encourages villagers to consume less water to reduce their bills and consequently improve water conservation consciousness via information transparency for water bills and the amount of consumed water.

Water is not only a product but a human right (Gleick 1998). Access to sufficient water for all has received wide attention from United Nations, WHO, and governments (Howard & Bartram 2003; Langford 2005). In less developed countries, private companies are more likely to be operating water utilities (i.e. contracted-out management) due to the limitation in financial support (Bakker 2007). The preceding discussion indicates that a rural water utility is costly to operate. Owing to the low water bills in rural areas, private companies operate small networks and may face low profits without subsidy. We found that for a water utility with a low population and a few subsidies, a private contract leads to IWS. Decreased staff and time costs through IWS (i.e. limited hours supply a day) are preferred for a private contracted utility. Hence, for a low population, absent of subsidies, using a contracted water utility may threaten the continuity of water supply.

This study conducts a quantitative analysis on the influence of institutional arrangements, economic support factors, and community context on the water supply modes of a village-scaled water utility in rural China. Three configurations based on economy, villager participation, and water saving lead to the CWS mode. Configurations lacking meter-based billing, subsidies, and water saving are the main reasons for IWS. This study highlights the importance of government subsidy, villager participation, water-saving programmes, and meter-based billing for the configurations towards improving the water supply mode in less developed countries. The following community-based policies and interventions are recommended to promote a sustainable and safe water supply in rural areas: (1) participation of local residents in water supply management in a small or unmetered billing water utility; (2) an emphasis on water conservation practices in village council operated utilities that lack subsidies; and (3) meter-based billing, subsidy by local authorities, and information sharing between users and contractors (or managers).

This study has certain limitations. Firstly, the cases were collected and gathered for only 38 village-level supply systems (20 CWS systems and 18 IWS systems). Although a flow procedure of case selection (i.e. identification, refining, and case selection) was adopted to maximise the representativeness of the selected cases, the small sample size may limit the generalisability of the findings. The total explanatory powers of the pathways of CWS mode and IWS mode are 76.8 and 87.1%, respectively, implying that a few underlying pathways can lead to CWS (23.2%) and IWS (12.9%) remain undiscovered. Further research can divide the research sites into several subregions on the basis of the differentiation of geography and socioeconomic characteristics. Case selection is based on the features of each subregion to improve the representativeness and coverage of samples. Secondly, this research examined only eight contextual factors of water supply mode. Although the contextual factors can represent the main variances in water supply mode, other factors, such as geography, water resource endowment, and household possessions may affect water supply mode and must be investigated. Future research needs to consider these additional factors to explain the configurations of water supply mode. Thirdly, this study focused on the cases of water utility scale and overlooked the analysis at household and individual villager scales. Household-scale qualitative data of behaviour observations and interviews for domestic water resource management are also valuable and must be further explored. Future work combining the regression-based method and fsQCA will be beneficial to discovering casual factors and public policy.

This work was supported by the National Natural Science Foundation of China (Grant No. 42171299), Program for Innovative Research Team (in Philosophy and Social Science) at the University of Henan Province (2022-CXTD-02; 2021-CXTD-08), Scientific and Technological Innovation Team of Universities in Henan Province (Grant No. 22IRTSTHN008), and the Nature Foundation of Henan Province (222300420172).

Special thanks to Dr Stacy M. Langsdale and three anonymous reviewers for their help and precious comments to revise and improve our paper.

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

The authors declare there is no conflict.