Household water insecurities and willingness to pay in Karachi

Ensuring equitable and sustainable distribution of clean water to households remains a significant challenge in developing countries. Universal access to safe and affordable drinking water is crucial for workforce health and productivity, making effective urban water management indispensable for economic development (Beckerman, 1992; Weststrate et al., 2018; Fukuda et al., 2019).

However, the combined forces of rapid urbanization, climate change, and weak water governance in many South Asian megacities have resulted in severe water stress, disproportionately impacting poor and marginalized communities (Beard & Mitlin, 2021). The poor quality of service provided by water utilities and resulting urban water insecurity hinder economic growth and exacerbate existing socioeconomic disparities (Enqvist et al., 2020).

In numerous cities, low-income households not only bear higher relative costs for procuring water (Adams, 2018; Rowling, 2019) but also face elevated health risks due to disparities in access to clean water (McDonald & Grineski, 2012; Bun et al., 2021). For example, Morales-Novelo et al. (2018) show that drinking water subsidies¹ in Mexico City disproportionately benefit high-income households, incentivizing wasteful water usage. Additionally, low-income households exhibit a higher willingness to pay (WTP) for improved water services when considered as a percentage of their corresponding income.

Karachi, one of the world's largest cities with a population exceeding 16 million, grapples with severe water scarcity, failing to meet 45% of its water needs (Toppa, 2016). Chronic issues related to planning and operation have led to revenue shortfalls and a lack of improvements in the piped water system. Despite hosting 60% of Pakistan's industry and contributing 12–15% annually to the national gross domestic product, Karachi suffers from inefficient water infrastructure, resulting in the loss of 30% of its domestic water supply due to theft or leakages (Akbar et al., 2021). The failure to accommodate the growing population's needs has exacerbated water scarcity, especially during the summer season. Additionally, inadequate maintenance of the aging piped water infrastructure has resulted in sewage mixing, posing significant public health risks (Kaleem, 2018). The poor quality of piped water is associated with approximately 30,000 annual deaths in the city (Pappas, 2011).

The disparities in access to piped water further compound existing socioeconomic inequalities, with households in low-income informal settlements spending nearly 10–20% of their monthly incomes to procure water from alternative non-piped sources (Khan & Arshad, 2022). The time spent procuring water also has implications for lost hours that could otherwise be dedicated to wage labor.

The Karachi Water and Sewerage Board (KWSB) is the sole utility responsible for managing water and sewerage services in Karachi. Years of mismanagement and political interference have left KWSB in a precarious financial situation (World Bank, 2018). It heavily relies on substantial provincial subsidies, accounting for nearly 35% of its total receipts, to cover operational expenses. The utility's weak financial state hampers its ability to undertake water supply augmentation or improve water infrastructure without support from the federal government or international donor agencies.²

KWSB's existing tariff structure and water resource management practices lack fairness and efficiency, leading to inequities and low billing collection rates. Tariff rates are extremely low compared to alternate water sources; on a volumetric basis, piped water rates, offered by KWSB, are between 10 and 100 times cheaper than water supplied through tankers (Khan & Arshad, 2022). These inefficiencies limit KWSB's ability to invest in water sources and maintenance, resulting in unreliable water supply. While affluent households can cope, less affluent ones cannot, leading to non-payment of bills and worsening service.

The literature on urban water infrastructure in the Global South suggests that households' WTP for improved water services can promote efficient and equitable water pricing, which is essential for sustainability in the water sector (Whittington et al. 1991; Venkatachalam, 2006; Vásquez, 2011; Vásquez & Franceschi, 2013). Household WTP is determined using the contingent valuation (CV) method, which constructs a hypothetical market scenario through various question formats. Respondents express their WTP or vote on proposed policies at specific price points. Several studies have highlighted the strong willingness of households to pay higher water tariffs in exchange for improved water services. For instance, Vásquez et al. (2009) used a random sample of 400 households in Parral, Mexico, to show that households were willing to allocate up to 7.5% of their median monthly incomes for safe and dependable drinking water. Similarly, Vásquez & Franceshi (2013) used a sample of 690 households and demonstrated that the total WTP for enhanced water services exceeded 8% of the reported median income, with variations depending on the service provider. In a separate investigation, Kwak et al. (2013) used data from 400 households in Pusan, South Korea, and identified a mean WTP for improved water services amounting to USD $2.2. Furthermore, Venkatachalam (2006) surveyed 206 households in Tamil Nadu to assess the WTP for enhanced tap water, revealing that individuals belonging to lower-caste groups exhibited a higher WTP for both individual connections and public taps. More recently, Ahsan et al. (2021) conducted a choice experiment involving 115 households in Khulna City, Bangladesh, and found that the total WTP of these households was approximately USD $2.87. This implies a readiness among respondents to invest in the enhancement of various water supply attributes, including water quality, regularity of supply, water pressure in taps, and water filtering. Van Houtven et al. (2017) offer a comprehensive analysis of previous studies that employed this method to estimate WTP and assess households' preferences for water quality improvements.

This study contributes to the existing literature from the Global South by offering an analysis of water access, payment behaviors, and WTP for improved water services in Karachi. It also investigates disparities in access to clean water, the associated economic costs, and health risks faced by low-income households. It provides valuable insights into patterns of marginalization and exclusion and offers policy implications for achieving equitable access to clean water in similar cities.

The CV method is a useful approach for assessing the value of non-market goods or services, such as public water resources, environmental quality, and wildlife preservation, where capturing consumers' revealed preferences through market prices is a challenge (Whittington et al., 1991; Carlsson et al., 2011). The CV method, based on a survey, elicits consumers' stated preferences, typically in the form of WTP, for any desired changes in the given good or service, such as improvements in water systems (Vásquez et al., 2009). Several studies have applied CV methods to elicit households' WTP for improved water services as a way to reduce the mismatch between offered services and consumers' expectations and preferences (Casey et al., 2006; Lema et al., 2012; Ahmed et al., 2022).

CV methods can be implemented in different formats. For instance, some researchers have utilized open-ended (OE) survey questions to elicit WTP (Carlsson et al., 2011; Shi et al., 2014), while others have predominantly relied on close-ended questions with dichotomous responses (yes/no) (e.g., Chatterjee et al., 2017). Additionally, some studies have incorporated a combination of both question formats (e.g., Ndebele & Forgie, 2017; Gordillo et al., 2019).

In the OE question format, respondents are asked to freely state the monetary value they are willing to pay for the desired change in a good or service. However, the OE format has been criticized as a weak elicitation method in scenarios where respondents must assign a monetary value to a hypothetical public good or service (Arrow et al., 1993). The OE format may fail to provide reliable valuation estimates, especially when respondents lack awareness about the good or service or when such valuations are uncommon (Reaves et al., 1999). In such cases, the dichotomous-choice (DC) format is recommended as an alternative (Arrow et al., 1993). However, when used alone, the DC format also has its limitations (see Bateman et al., 2001).

In the DC format, respondents are presented with monetary values (bids) for the desired change and may choose to accept or reject each bid. The game proceeds in rounds, with higher or lower bids shown based on the respondent's previous decision. This format allows for estimating lower and upper bounds on respondents' WTP and can be employed in single-bounded or multiple-bounded formats (Cooper et al., 2002).

Despite the usefulness of CV methods in capturing consumers' valuation of non-market goods or services, these estimates may suffer from various biases, including starting point bias and hypothetical bias (HB; Murphy & Stevens, 2004; Morrison & Brown, 2009). Starting point bias occurs when respondents strategically alter subsequent responses based on the initial value shown (Herriges & Shogren, 1996) and can also arise due to ‘yea-saying’ behavior, where respondents agree to questions regardless of their actual preferences (Yeung et al., 2006). To address ‘yea-saying,’ previous studies recommend including follow-up questions or informal conversations to filter out non-serious responses (Blumenschein et al., 2007; Penn & Hu, 2018). Moreover, the National Oceanographic and Atmospheric Association (NOAA) CV Panel recommends using the DC format as it minimizes the role of strategic behavior among individuals and can yield reliable estimates in large sample sizes due to its resemblance to everyday market transactions (Arrow et al., 1993).

Another important critique of the CV method is the potential presence of HB, which refers to the difference between a respondent's stated WTP and their actual WTP in real market transactions. Since CV methods involve hypothetical scenarios about payments for and the provision of goods or services, it is challenging to determine how respondents would behave in the real world compared to a hypothetical scenario (Ajzen et al., 2004).

To address these concerns, we have implemented various strategies to minimize biases in our study design. These strategies include follow-up informal conversations with respondents, a suitable questionnaire format, and enumerator training. Our study primarily focuses on improvements in piped water services, which have direct use values, thus limiting the HB. We also conduct follow-up informal conversations to reduce ‘yea-saying’ and ‘hypothetical bias’ by assessing respondents' confidence in their quoted bids. Bids for improved piped water services are designed to reflect consumers' rational behavior in response to income constraints and cost considerations in non-piped water markets, mitigating starting point bias concerns. Additionally, our survey design incorporates both OE and DC question formats to minimize the role of ‘starting point’ bias. Enumerators are trained to elicit an OE bid from respondents, followed by a DC bid based on the value of the OE bid, helping to minimize starting point bias and strategic behavior (we explain this further in the appendix under Survey Instrument Design).

Sampling bias arises when the sample fails to accurately represent the population due to systematic errors in the sampling procedures, potentially leading to results that do not reflect the true population characteristics (Sharma, 2017). This bias can occur if the strata are improperly defined or if certain groups are either underrepresented or overrepresented in the sample.

To mitigate the risk of sampling bias, we implemented a series of carefully considered steps. We employed a well-designed stratified random sampling procedure to ensure representativeness (Cochran, 1977; Sharma, 2017). In this method, we divided the Karachi population into district-wise strata and randomly selected one town from each stratum. Within each selected town, we then randomly chose five union councils (UCs) and conducted surveys among six to nine households from each UC. This approach ensured that individual observations within each stratum were drawn using simple random sampling, guaranteeing that each element in the population had an equal chance of being included in the sample (Cochran, 1977; Rahi, 2017).

By employing the randomized procedures within our stratified random sampling approach, we aimed to capture a representative cross-section of characteristics related to households' access to water in Karachi, excluding cantonment areas. We sampled households from KWSB's service area and excluded cantonment areas that are subject to distinct administrative procedures.

Using this formula, we determined the relative sample sizes for the strata, considering the district-wise population distribution in Karachi, and made proportional adjustments for higher confidence and increased precision. Table 1 presents the total number of surveys conducted within each stratum (district/town) in each survey round.

Previous studies with similar sample sizes have explored residents' WTP for clean water services in the Global South, although city-wide studies are limited. For instance, Jalilov (2017) conducted 240 structured interviews to gauge residents' WTP in Manila. Pattanayak et al. (2005) analyzed 1,500 households in Kathmandu, Nepal, finding coping costs associated with poor water infrastructure could reach 3 USD per month, equivalent to 1% of current incomes. These costs exceeded the monthly water bills but were lower than estimates of WTP for improved services. They also identified a statistical correlation between coping costs and WTP, which is also influenced by various household characteristics.

In another study, Jianjun et al. (2016) investigated 168 households in Songzi City, Hubei Province, China, and determined that the mean WTP for enhanced drinking water quality was 0.3% of total household income. Factors such as health considerations, household income, and the perceived importance of water quality exhibited positive associations with household WTP. Additional information and discussions about relevant studies, particularly those from the Global South, can be found in Table 7 in the appendix.

Our survey questionnaire included questions on socioeconomic and demographic characteristics, household reliance and expenditures on piped vs non-piped water sources for drinking and household activities, and expenditures on health and non-water utilities, including electricity, gas, Internet, and cable (please see the Supplementary Files for the dataset, description of the tools and data, and the questionnaire). A key section in the survey asked respondents about satisfaction with, and WTP for, different attributes of piped water supply (pressure and sufficiency, quality, convenience, and reliability of supply schedule). The composite term Pressure and Sufficiency captures the flow rate and the volume of water supplied. Quality of water refers to whether the water supplied is safe for drinking and is determined by respondents through the presence of visible impurities as well as through smell and taste. Convenience is defined in terms of irregularity of supply timings; a supply schedule is deemed inconvenient if it coincides with resting hours (12–5 AM). Reliability is a measure of the consistency of the water supply schedule.

We elicit WTP for each aspect of piped water through both OE and DC formats. This helps reduce starting point bias with choosing any specific opening bid (Loomis & McTernan, 2014). Respondents indicating dissatisfaction with any attributes of the piped water supply are asked questions about WTP for improvements. In the OE format, we ask respondents the following question: ‘How much do you think you or anyone from your household would be willing to pay per month additionally to receive improved < attribute of water service > ?’ Respondents report opening bids through an OE response. For respondents not willing to pay anything, enumerators would record the bid amount as zero. Trained field enumerators used small talk to ascertain the authenticity of the response, ensuring consistency and certainty of responses. Only households with a piped water connection were asked to play the bidding game (please see the appendix and survey description in the Supplementary Material for our survey tools, design, and description).

In the DC format, the surveyors read the following prompt for the respondents: ‘While the exact cost of improved water services is not known with certainty, we expect them to range from PKR 500 to 1,500’. If respondents answer yes to the lowest bid (PKR 500), they are then presented with higher bids, in increments of PKR 500, up to PKR 1500. Compared to the OE format in which respondents can freely quote any amount, the DC format restricts bid amounts by making them conditional on the OE response given earlier. For example, a respondent initially quoting PKR 400 in the OE format would be presented with a bid of PKR 500 to accept or reject in the DC format, and so on. On the other hand, a respondent quoting PKR 600 in the OE format would be presented with a bid of PKR 1,000 in the DC format.

To determine the extent of spatial and socioeconomic inequities in the access to water, we look at the descriptive statistics of our sample households. To better understand household behavior concerning water bills, we perform a multivariate regression analysis on the likelihood of bill payment while controlling for spatial (town) and socioeconomic inequities (income). For WTP, we look at the mean WTP for each water service (pressure, quality, convenience, and reliability) across income groups.

Table 2 presents descriptive statistics for the key variables in this study. The mean WTP for improvements in pressure and sufficiency is PKR 839, with a median of PKR 500. The mean WTP for enhanced water quality is PKR 842, also with a median of PKR 500. The mean WTP for convenience and reliability of the water supply schedule is lower, at PKR 572 and 539, respectively, while the medians for both remain at PKR 500. The combined WTP for improvements in all four service attributes is PKR 1,149, with a median of PKR 1,000, representing nearly 3% of the reported median income. The median WTP for safe and reliable access to water aligns with previous literature in the Asian context (Pattanayak et al., 2005; Kwak et al., 2013; Gunatilake & Tachiri, 2014; Jianjun et al., 2016). In Bangladesh, Pattanayak et al. (2005) found a lower estimate at 1% of the mean income. In terms of demographic and socioeconomic characteristics, respondents are almost evenly distributed among males (49%) and females (51%). The average age of the respondents is 42 years. The median size of the houses is 80 square yards. The average years of schooling for any member of the household are above 12 years, indicating the completion of higher secondary school. In the sample, nearly half of piped households have at least one member with tertiary education, compared to just over a fifth of non-piped households. Over 77% of the households have voted in provincial or national elections, suggesting active engagement in public service provision.

Furthermore, we created income categories similar to those employed in the World Bank report (World Bank, 2018). We created income categories based on Pakistan's Household Integrated Economic Survey (HIES) 2018–2019. There were three categories: <45,000, 45,000–75,000, and > 75,000. By utilizing these income categories, we examined disparities in piped water access across various neighborhoods. As illustrated in Figure 2, low-income neighborhoods in Malir, Korangi, Orangi, and Lyari are disproportionately more likely to lack piped water supply (0 hours per week; represented in black). Conversely, areas with improved access to water (6–42 hours per week or more than 42 hours per week) are predominantly situated in higher-income neighborhoods. This highlights that the water resource management system in Karachi is shaped by socioeconomic disparities among neighborhoods, with favorable service provision in wealthier areas. To explore further, we examine pipeline coverage at the town level.

In Panel B of Figure 4, we observe that 76.7% of low-income households experience sewage mixing regularly, compared to 55.4% of high-income households. Overall, nearly 70% of surveyed households report sewage mixing in their piped water supply. This high frequency is a grave concern, particularly in light of the significant number of deaths in Karachi attributed to contaminated water (30,000 annual deaths) (Pappas, 2011).

Our data on water supply duration reveals a recurring pattern of low pressure and leakages, particularly prevalent in low-income neighborhoods. This issue is demonstrated in Table 3, where Gulshan-e-Iqbal town boasts the highest supply with a median of 2,520 minutes per week, while Malir and Orangi towns grapple with severe shortages (0 and 15 min, respectively). These findings are consistent with previous research, indicating that Malir and West districts receive only 30–57% of their water quota, in contrast to other districts in upper-income neighborhoods, which can receive up to 60–100% (Janjua et al., 2021). Karachi's aging water infrastructure, now over 30 years old, exacerbates problems such as leakages, theft, and sewage contamination in the piped water supply. Operational inefficiencies and financial challenges faced by KWSB further impede regular maintenance and service provision to low-income neighborhoods.

Billing and recovery of water bills pose significant challenges in Karachi. Our survey revealed that, of the 818 households with a pipeline connection, only 58% of the households receive a monthly bill from KWSB. Among the households receiving a water bill, several interesting behaviors are observed. Nearly 56 households, primarily in Korangi and Malir towns, receive and pay their water bills despite not receiving any supply of water in their pipelines. When asked about their rationale for paying despite no service provision, these households explained that maintaining a pipeline connection can have a positive impact on their property values. They hold the hope of future service provision, which they believe could prove advantageous if they decide to sell their house. On the other hand, 77 households in our dataset can be categorized as defaulters; they receive piped water and monthly water bills but do not pay those bills. Overall, only three-quarters of KWSB's registered retail customers pay their bills.

Previous research suggests that households facing irregular or no water supply may be less likely to make timely bill payments (Dutta et al., 2005). In this study, we test this hypothesis: whether the availability of piped water significantly influences bill payments.

Table 4 presents the results of the logistic regression model aimed at understanding the factors influencing household water bill payments in Karachi.³ The table includes odds ratios and marginal effects, with standard errors in parentheses. First, being male is associated with 1.32 times higher odds of paying the water bill, indicating a gender difference in bill payment behavior. The total number of household members and years of schooling have minor effects on bill payment, with marginal effects close to zero. Notably, the variable water unavailability demonstrates a substantial impact, as respondents facing water unavailability have significantly lower odds of paying their bills, with an odds ratio of 0.489 and a substantial negative marginal effect of −0.110; it indicates that experiencing water unavailability is associated with an 11 percentage point decrease in the probability of paying the water bill. Controlling for the weekly supply of water (not shown in the table), the relationship remains statistically significant and strong at 12 percentage points.

The results also indicate the significance of household location in influencing water bill payments. Residents of Liaquatabad have a 14.3 percentage point higher probability of paying the water bill compared to residents of Malir Town. Similarly, the marginal effect of Gulshan-e-Iqbal (Karachi East) is 0.218, which indicates that residents of Gulshan-e-Iqbal have a 21.8 percentage point higher probability of bill payment compared to residents of Malir Town.

Our analysis of household water bill payments provides a valuable link to the broader analysis of WTP for improved water services. The factors influencing bill payment behavior, such as water supply schedule and location-based disparities, directly impact the willingness of households to invest in improved water services.

The median WTP for improvement in each attribute of piped water service (pressure and sufficiency, quality, convenience, and reliability) is PKR 500 and that for combined improvements in all four attributes is PKR 1,000. However, WTP varies substantially by household income (see Table 6). Based on the data from Pakistan's Household Income and Expenditure Survey (HIES) of 2018–2019, households are categorized into the following three income groups: low-income (less than PKR 45,000), middle-income (between PKR 45,000 and PKR 75,000), and high-income (more than PKR 75,000).⁴

Table 5 shows that the median WTP per month is PKR 500 for low-income households, PKR 1,000 for middle-income households, and PKR 1,700 for high-income households (median KWSB monthly bill is PKR 390). For comparison, previous findings suggest a mean WTP for safe piped water ranging between PKR 400 and 500 in Islamabad (Ahmed et al., 2022). In our sample, the monthly water bills average PKR 642 for low-income households, PKR 578 for middle-income households, and PKR 745 for high-income households.

In Table 6, Kruskal–Wallis tests were conducted to assess how WTP for the improvement of different aspects of piped water supply varied among households from various income brackets. The results indicate significant differences in OE bids for pressure & sufficiency, water quality, convenience of water supply, and the combined bid across income groups all at a 1% level of significance. The bids for improving the reliability of the water supply variable also exhibited a significant difference, albeit at a 10% level of significance. In the context of non-parametric tests like the Kruskal–Wallis test, tied data can impact rank-based calculations; hence, adjustments were made to accommodate these tied values. The results of Dunn's pairwise comparisons of bids by income category provide insights into the variations in WTP for the improvement of different aspects of piped water supply among households from different income brackets. Significant differences were observed in multiple categories. For pressure & sufficiency and water quality, Group 1 (with incomes below PKR 45,000) demonstrated highly significant differences when compared to Group 3 (incomes above PKR 75,000), while Group 2 (incomes between PKR 45,000 and 75,000) also showed significant differences in the case of water quality. In contrast, the convenience of water supply exhibited significant disparities between Group 1 and Group 3, with a moderately significant difference between Group 2 and Group 3. The reliability of water supply variable did not reveal significant differences among the income groups. However, the combined bid showed significant disparities in all pairings, with highly significant differences between Group 1 and Group 3, as well as Group 2 and Group 3. Overall, our findings suggest that income disparities have a notable impact on households' WTP for the enhancement of various aspects of piped water supply, highlighting the importance of income-sensitive interventions in water supply planning.

These findings align with the existing literature on improved urban water services in South Asia (Majumdar & Gupta, 2007; Dutta & Verma, 2009), Southeast Asia (Rietveld et al., 2000; Lee et al., 2016), Central Asia (Tussupova et al., 2015), and Sub-Saharan Africa (Nti et al., 2020). Water utilities that have implemented tailored water pricing policies have been successful in converting higher WTP into increased revenue. Overall, in many developing countries, there is a shift away from traditional, non-volumetric tariff structures, as seen in Karachi, toward tariffs that include a volumetric tariff structure and variable charges sensitive to various income brackets. These measures provide consumers with incentives to use water efficiently (Shen & Reddy, 2016). Nonetheless, there is a growing consensus among scholars and activists on the need to protect low-income households. These protections can be achieved through measures such as subsidizing initial connection costs, legalizing small-scale water vending, or implementing increasing block tariffs with equitable block sizes sensitive to household socioeconomic constraints (Dinar, 1998; Whittington, 2003; Soto Rios et al., 2018).

Our findings also contribute to the existing literature aimed at structuring consumer-oriented pricing models, especially in urban water supply management. Understanding WTP is vital for aligning consumer valuations with water service improvements when discrepancies exist between their valuation and utility tariff structures (Chatterjee et al., 2017). Consistent WTP estimates and income-sensitive interventions can aid public utilities like KWSB in prioritizing service enhancements to ensure financial sustainability. Presently, KWSB's tariff structure for residential customers, based on property size, is both inefficient and unsustainable, discouraging water conservation and disadvantaging low-income households. This inefficiency is linked to low tariff collection rates, limiting KWSB's ability to expand services to underserved neighborhoods. In areas with lower-income households, there is a pressing need for income-sensitive interventions aimed at improving the pressure, sufficiency, and water quality of piped water supply. While the Kruskal–Wallis test did not reveal significant differences in the reliability of water supply among income groups, the importance of ensuring and maintaining the reliability of water supply remains significant for all income brackets to guarantee consistent service.

We present a case study from a major city in the Global South, Karachi, to highlight patterns of marginalization and exclusion in access to piped water connections. Findings in this study reveal that households in low-income neighborhoods have limited or no access to piped water connections and experience more frequent water shortages and sewage mixing compared to households in high-income areas.

Collecting survey data from 990 households across Karachi, this study primarily aims to inform policy efforts toward more sustainable and equitable urban water management. It quantifies disparities in piped water, highlights differences in household bill payment behaviors, and investigates the household WTP for various attributes of piped water services. Using Karachi as a case study, our findings offer key insights into urban water policy in the Global South.

Our analysis underscores the disparity in pipeline coverage in Karachi, with the lowest coverage observed in low-income areas like Korangi (71%) and the highest in upper-income areas, notably Gulshan-e-Iqbal (99%). Low-income households also bear a disproportionate burden of sewage mixing, with over 76% of them experiencing this issue frequently or occasionally, compared to 55% of high-income households.

We find that households experiencing the unavailability of water in the past month have an 11% lower likelihood of paying their bills compared to those who received piped water over the same period. Our study highlights that KWSB's service provision remains unsatisfactory in Karachi. Households have developed alternate coping strategies to overcome chronic piped water issues, which are often more expensive than the tariffs charged by KWSB. Nevertheless, households are willing to pay additional amounts for water service improvements. The median WTP is PKR 500 for improvements in any aspect of water (pressure, quality, convenience, and reliability). The combined WTP for improvements in all four aspects is PKR 1,000, with substantial income-based variation. The median monthly WTP is PKR 500 for low-income households, PKR 1,000 for middle-income households, and PKR 1,700 for high-income households. The combined WTP for improvement in all the aspects of water supply is nearly 3% of the median income, which aligns with similar estimates obtained from urban centers in South Asia.

Our analysis carries significant policy implications. First, the existing low level of service, particularly concerning the water supply schedule, which includes aspects like water supply frequency and duration, may discourage households from fulfilling their water bill obligations. This underscores the importance of addressing and improving the reliability and availability of water supply services to encourage timely bill payments. Second, our analysis indicates that there may be location-based differences in bill payment likelihood. This disparity may be linked to inconsistent service delivery levels among regional offices, as evidenced by the significant differences in water supply across different towns. These findings highlight the importance of addressing regional disparities in service delivery and promoting more equitable access to quality water supply services across different localities in Karachi. Additionally, the WTP estimates for improved water services offer guidance for designing consumer-centric cost structures for piped water services. In areas with lower-income households, implementing income-sensitive interventions becomes particularly crucial for enhancing the pressure, sufficiency, and water quality of piped water supply. To improve convenience, such as accessibility and ease of accessing water supply, strategic interventions like infrastructure development, upgrades to the water supply system, and enhanced maintenance practices can greatly benefit lower-income households, making water supply more accessible and convenient for them. Notably, the combined bid for enhancing various aspects of piped water supply displayed significant disparities across income groups. Therefore, policymakers should take into account the distinct needs and preferences of different income groups when designing and implementing improvements to various facets of piped water supply. Tailored interventions can promote equitable access to high-quality water supply services and address disparities in WTP among households from diverse income brackets. In terms of specific measures, implementing community-specific water quality improvement measures, particularly in areas prone to contamination or low-quality water sources, is needed. Additionally, community involvement in decision-making processes related to water services is essential. The establishment of local water management committees can contribute to promoting a culture of water conservation through community-led initiatives and education programs.

This research was supported by a grant (#13659) from the Higher Education Commission of Pakistan (HEC) as part of the National Research Program for Universities (NRPU).

S.K. expresses gratitude to the 2023 International Behavioral Public Policy Conference participants at the University of North Carolina at Chapel Hill and colleagues at the University of Washington for their valuable feedback. The authors also appreciate the insightful feedback and suggestions provided by the reviewers. We are also thankful for the research assistance from students at Habib University, with special recognition for Muhammad Aqib Yousuf for excellent research support and Ali Arshad, Fariha Batool, Zayaan Delawalla, Aimen Imtiaz, Khudaija Reza, and Rida Rehan Chughtai for their pivotal roles in fieldwork and data collection.

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

The authors declare there is no conflict.