Estimating Household Water Demand and Affordability Under Intermittent Supply: An Econometric Analysis with a Water–Energy Nexus Perspective for Pimpri-Chinchwad, India

Abstract

Urban water utilities in rapidly developing regions face growing challenges in ensuring continuous supply. Intermittent public water supply leads to unreliable and inequitable access, compelling households to adopt energy-intensive coping strategies. This creates a nexus between water and energy demand at the household level. Few econometric analyses of household water demand have explicitly addressed this demand-side nexus in developing regions. Using survey data from the city of Pimpri-Chinchwad, India, where intermittent water supply is prevalent, we analyze household expenditures related to water access and estimate a piped water demand function with a Discrete-Continuous Choice model. We find that electricity expenditures for accessing water exceed water bills for approximately one-third of households. Including these costs in affordability calculations reveals hidden financial burdens, particularly for middle-income households. Water and electricity prices, income, and household size significantly influence water demand, with an income elasticity of 0.177 and water price elasticities ranging from 0 to −0.876. The cross-price elasticity of −0.097 indicates weak complementarity between electricity and piped water, suggesting electricity price changes do affect water use but are insufficient to drive substantial behavioral shifts. Targeted price increases in high-consumption blocks are more effective at curbing overuse, while simultaneous increases in water and electricity prices may heighten household vulnerability. These findings highlight the need for integrated, nexus-aware demand management strategies, particularly in regions with intermittent supply.

1. Introduction

Many developing regions face growing challenges in urban water demand management, as rapid population growth and urban expansion exert increasing pressure on municipal water supply systems [1,2,3,4,5]. Increasing demand and aging or inadequate infrastructure often result in intermittent water supplies. While this is sometimes a temporary or cost-effective solution for utilities with limited resources [6,7], it frequently forces private households to adopt coping strategies, such as installing rooftop storage tanks or relying on private water tankers [5,8,9,10,11,12,13].

Equity issues are more prevalent in such settings. Since water supply hours are often distributed unevenly, poorer households may receive fewer hours of service or lack access to in-house piped connections [9,12,14]. The demand-side water–energy nexus at the household level further exacerbates these disparities. This nexus refers to the economic and behavioral linkages between water and energy use within households, where the consumption of one resource is directly or indirectly influenced by the consumption, price, or accessibility of the other [15]. Household decisions related to domestic activities and coping strategies, such as the use of appliances and energy-intensive water storage or treatment systems, all contribute to these interdependent demand patterns. Additional costs often arise as households cope with intermittent supply. For example, Cook et al. [16] found that the median total coping costs in rural Kenya exceed the average urban water bills in many cities in the United States, with costs such as storage and drinking water treatment adding to predominantly time-based costs. Similarly, Pattanayak et al. [17] estimated that household coping costs, including costs for pumping and water treatment, can be nearly twice as high as utility water bills in Kathmandu, Nepal. In a case study of urban areas in the Pune Metropolitan Region, India, Zhu et al. [9] identified a statistically significant correlation between water supply hours and household electricity demand, and found that the electricity used to access water can account for up to 27% of a household’s total electricity consumption on average, with even higher shares during droughts. These additional costs imply hidden financial burdens that households may face under intermittent water supply, and poorer households tend to be more affected [9,12,16,18]. Such burdens are often overlooked by standard affordability metrics, which focus only on direct water charges.

Adopting a water–energy nexus perspective can offer deeper insight into distributional factors in water demand management and decision-making, which is essential for ensuring universal and equitable water access [19,20]. While empirical estimates of water price elasticities suggest that households are often rationally ignorant of water prices [21,22], it remains uncertain whether they recognize or factor in indirect costs, particularly electricity costs associated with accessing water. These costs can be substantial in contexts with intermittent water supply, where electricity is frequently used for pumping and storage to ensure water availability [6,9,23]. This underscores the pressing need to understand the demand-side water–energy nexus at the household level and to address equity issues caused by intermittent water supplies in urban water demand management.

Although conceptually analyzed [15], quantitative empirical evidence on this demand-side nexus at the household level remains limited, particularly regarding how electricity expenditures related to coping with intermittent water supply influence demand for piped water. Few studies have estimated household-level cross-price elasticities between electricity and piped water, especially in developing regions. This results in a knowledge gap regarding the economic interlinkages in household resource use. The lack of empirical insight also limits the ability of utilities and policymakers to develop pricing and demand management strategies that account for the full financial burdens borne by households in contexts of intermittent water supply, which may undermine the equity and effectiveness of urban water management. To address these gaps, we conducted an empirical econometric analysis of household piped water demand in Pimpri-Chinchwad, a major industrial and residential center in India experiencing intermittent water supply. Based on an estimated water demand function that includes both water and electricity prices, our research is guided by three key questions: (1) How do electricity costs for accessing water under intermittent supply affect water affordability? (2) What is the cross-price elasticity between electricity and piped water at the household level? and (3) How do changes in water and electricity prices impact household piped water demand and households’ ability to access piped water for essential uses? By answering these questions, we aim to improve understanding of the demand-side water–energy nexus at the household level in regions with intermittent supply and support urban water management strategies that aim to ensure secure, affordable, and equitable water access in developing regions.

To this end, we analyze a sample of standard urban households in Pimpri-Chinchwad drawn from a comprehensive survey of the Pune Metropolitan Region in India. First, we calculate both standard and nexus-based water affordability ratios for a subsample with accurate water billing data to assess whether standard affordability measures adequately reflect the full financial burden households faced under intermittent water supply. We hypothesize that electricity expenditures related to accessing water are substantial and that including these costs in affordability metrics will reveal a higher incidence of affordability stress than standard methods suggest. We then estimate a household piped water demand function that includes water and electricity prices, income, and household size as explanatory variables using a Discrete-Continuous Choice (DCC) model. Given the widespread use of electric pumps to cope with supply interruptions, we expect piped water and electricity to function as economic complements, and therefore hypothesize a negative cross-price elasticity between them. Based on the demand function estimates, we further assess the impact of changes in water and electricity prices on household piped water demand through simulations of hypothetical price scenarios. The remainder of this paper is organized as follows: Section 2 outlines the case study context and methods for calculating water affordability, estimating demand, and analyzing price impacts in an explicit nexus setting. Section 3 presents results on water-related expenditures and affordability, estimates of household piped water demand, and simulation results on hypothetical price scenarios. Section 4 discusses the implications for urban water management. Section 5 concludes with the main findings.

2. Materials and Methods

2.1. Case Study Area

Pimpri-Chinchwad is part of the Pune Metropolitan Region, one of India’s largest and fastest-growing urban agglomerations. Located in western Maharashtra, the city covers an area of 181 km² [24] with a population exceeding 1.7 million [25]. Pimpri-Chinchwad is undergoing rapid urban expansion, resulting in significant urban sprawl and increasing pressure on municipal infrastructure, particularly the water supply system [1,26,27].

The intermittent piped water supply is widespread in Pimpri-Chinchwad. In the urban areas of the Pune Metropolitan Region, which includes both Pimpri-Chinchwad and its neighboring city, Pune, households typically receive piped water for about six hours per day, with a notable inequality in water access in terms of the supply durations [9]. To cope with water supply intermittency, many households invest in storage tanks and electric pumps to meet their daily water needs. There is also increased reliance on groundwater and private water tankers, especially during the summer months [9]. These coping strategies not only result in high electricity consumption to manage water problems, but also highlight further issues of inequality in water access, as wealthier households often receive longer supply hours and are also better able to afford the coping solutions. This context makes the region particularly relevant for examining the demand-side water–energy nexus at the household level, and for exploring how households are affected by the challenges of intermittent supply and inequality in water access.

Unlike the city of Pune, where domestic water is typically charged at a flat rate linked to property tax and most households lack functional water meters, Pimpri-Chinchwad has introduced water metering for the majority of urban households and adopted a volumetric pricing system [28,29]. Since 2018, the municipal corporation has employed an increasing block tariff (IBT) structure for domestic piped water, dividing consumption into four blocks with progressively higher rates for higher consumption levels. Each household receives the first 6 m³ of piped water per month for free, while higher rates apply for subsequent consumption blocks: between 6 and 22.5 m³, between 22.5 and 30 m³, and above 30 m³. These rates increase by 5% annually, as documented in municipal records [29] (pp. 14–15) and verified by actual water bills from 2019 to 2023, as presented in

Table 1. Domestic Piped Water Tariff Structure in Pimpri-Chinchwad from Financial Year 2019/20 to 2022/23, Rates in INR.

Consumption Block (m3)	Rates FY 2019/20	Rates FY 2020/21	Rates FY 2021/22	Rates FY 2022/23
0–6	0	0	0	0
6–15	4.2	4.41	4.63	4.86
15–22.5	4.2	4.41	4.63	4.86
22.5–30	8.4	8.82	9.26	9.72
above 30	12.6	13.23	13.89	14.58

. The extensive use of water meters and volumetric pricing in Pimpri-Chinchwad allows for in-depth analysis of household piped water demand and the estimation of a demand function. Therefore, it is an appropriate and revealing case study area to answer our research questions.

Generally speaking, water pricing serves multiple policy objectives, including cost recovery, economic efficiency, equity, and environmental sustainability, as widely recognized in both policy and academic literature [30,31,32]. In practice, achieving these objectives requires a thorough understanding of local economic and infrastructural conditions. As shown in Table 1, water prices in Pimpri-Chinchwad have increased modestly in recent years, with the annual 5% adjustment primarily offsetting inflation rather than driving meaningful rate hikes. However, the municipal corporation has explored more aggressive pricing strategies to strengthen demand management. For example, for the 2020/21 financial year, the municipal corporation proposed sharp tariff increases targeting households with piped water consumption exceeding 100 Liters per Capita per Day (LPCD) [33] (pp. 31–32). The proposal retained the free 6 m³ per month for households consuming less than 20 m³ monthly (approximately 135 LPCD), but suggested significantly higher rates for higher consumption blocks. Under the proposed structure, the rate for the 6–15 m³ block would rise from 4.2 to 8 Indian Rupees per m³ (INR, 1 INR equals 0.0142 USD in 2019), and the 15–20 m³ block would be charged at 40 INR per m³, nearly ten times the previous rate. Consumption above 30 m³ per month would be priced at 100 INR per m³, compared to the previous rates of 8.4 to 12.6 INR per m³. Additionally, households consuming more than 20 m³ per month would no longer receive the free 6 m³; instead, all consumption up to 15 m³ would be charged at 8 INR per m³. Importantly, this proposed tariff structure was never implemented, as confirmed by actual water bills from that period. Nevertheless, it indicates that significant price hikes may occur in Pimpri-Chinchwad, either to curb excessive demand or to address potential water crises caused by unexpected droughts, which emphasizes the potential policy relevance of our study for the case study area.

2.2. Data

We use data from the Pune Household Food–Water–Energy Nexus Consumption Survey, conducted in 2020 across the urban areas of the Pune Metropolitan Region. The complete survey dataset includes 1872 household observations from both formal and informal settlements in Pimpri-Chinchwad and Pune. As described in Section 2.1, domestic water in Pune operates under a flat-rate billing system linked to property taxes, meaning piped water charges are based on dwelling size rather than actual consumption. Similar conditions exist in most informal settlements throughout Pimpri-Chinchwad, where functional water meters and regular billing systems are typically unavailable. Under these circumstances, the relationship between observed household water use and the price paid becomes negligible, which undermines reliable estimations of price responsiveness. Therefore, for the econometric analysis in this study, we focus on the standard urban household observations in Pimpri-Chinchwad, where water metering and billing systems operate more reliably, and exclude those residing in informal settlements, resulting in a sample of 568 observations.

Of the 568 standard urban households surveyed in Pimpri-Chinchwad, 92.96% have a metered piped water connection, and 96.30% reported paying for piped water. However, among those who paid for piped water, only 60.15% of households stated that they were billed on a volumetric basis, and 222 households provided the exact amount of water consumption from their water bills. In addition, 41.55% of the 568 households provided at least one full month of electricity consumption from their bills. This leads to a total of 170 households with accurate data on both water and electricity consumption. For the estimation of the demand function, we exclude households with more than 20 members or with a water consumption of more than 1000 LPCD to minimize the impact of outliers. This results in a demand estimation sample of 160 valid observations. To validate the representativeness of the demand estimation subsample, we compare its key characteristics with those of the full sample. As shown in

Table 2. Descriptive Statistics for the Full Sample of Standard Urban Households in Pimpri-Chinchwad (n = 568) and the Subsample for Demand Estimation (n = 160).

Statistic	Full Sample of Standard Urban Households in Pimpri-Chinchwad			Subsample for Demand Estimation
	N	Mean	St. Dev.	N	Mean	St. Dev.	Min	Max
Household Income (INR/Month)	559	40,503.58	36,160.35	160	39,518.75	45,257.61	3000	500,000
Household Size (Count)	568	4.94	2.89	160	5.02	2.34	1	16
Piped Water Supply Hours (Hour/Day)	372	3.63	5.16	104	3.76	6.03	0.50	24.00
Use of Electric Water Pump (Dummy)	568	0.52	0.50	160	0.65	0.48	0	1
Monthly Household Piped Water Consumption (m3/Month)	-			160	24.47	13.41	2.40	84.61
Monthly Household Electricity Consumption (kWh/Month)	-			160	121.51	124.34	0.80	1304.25
Marginal Price for Water (INR/m3)	-			160	7.01	3.70	0.00	12.60
Marginal Price for Electricity (INR/kWh)	-			160	7.80	2.70	5.45	17.06

, the subsample’s distributions of income, household size, piped water supply hours, and pump use closely mirror those of the full sample, indicating that the subsample remains broadly representative of the overall population of standard urban households in Pimpri-Chinchwad.

2.3. Methods

2.3.1. Affordability Calculation

To answer the first research question regarding water affordability assessments, we use the affordability ratio method. The concept of affordability ratios originated in housing studies, where they were developed to evaluate the financial burden of housing costs relative to household income [34]. This approach was later adapted for use in the water sector and is now widely used to assess whether households can afford water and sanitation services while meeting other basic needs [35,36,37]. Various calculation methods have been proposed over time for calculating affordability. These include the conventional affordability ratio that measures the share of household expenditures on water and sewer services relative to income, the potential affordability approach that considers expenditures for a specified quantity of water relative to income, the residual income approach that focuses on whether households have sufficient income left over after paying for an essential amount of water for basic standard of living, the capital expenditure relative to income that accounts for one-time costs of capital expenses relative to income, etc. [37,38,39]. Among these approaches, the conventional affordability ratio remains one of the most commonly used due to its simplicity and moderate data requirements, particularly in policy and regulatory contexts [38,40,41]. While approaches such as the residual income method offer advantages by capturing a broader perspective of household financial capacity, they require detailed data on household expenditures beyond water and minimum living costs, which are not available in our survey dataset. Therefore, for this case study, we employ the conventional affordability ratio method, calculated as shown in Equation (1) below. It is important to acknowledge that this method may underestimate affordability burdens because it does not account for other essential household needs; therefore, our results likely represent a conservative estimate of water affordability challenges.

$$AR={\displaystyle \frac{Water Expense}{Household Income}}\times 100\%$$

(1)

With this calculation, water is considered affordable if household water expenses do not exceed a specified percentage of income. Thresholds vary in the literature depending on whether the calculation includes only water or both water and sewer costs. When considering water costs alone, the affordability threshold typically ranges from 2% to 5% [39,40,42,43]. In Pimpri-Chinchwad, water bills cover only piped water provision and do not include sewerage or wastewater treatment costs, as a citywide, well-functioning sewerage system has not yet been fully established. Therefore, we use a threshold of 2.5%, which has been frequently adopted in recent studies [40], to define water affordability.

Water affordability is often measured at the community or regional level, using the average cost of water services and the median household income of the community or region. Recent studies also suggest using the 20th income percentile, along with complementary metrics such as converting water costs into the number of minimum-wage labor hours, to more accurately represent affordability characteristics [38,40]. However, since households at different income levels have varying capacities to cope with costs, in particular, the additional costs of coping with intermittent water supply, calculating affordability at the household level provides more detailed insights than relying solely on regional averages or percentiles [19]. Accordingly, we calculate household-level conventional affordability ratios based on the demand estimation subsample presented in Section 2.2, which contains accurate household-level water billing data.

Specifically, we calculate two variants of the conventional affordability ratio at the household level: (i) The standard affordability ratio, where the water expense in Equation (1) is the actual amount of water bill paid by households; (ii) The nexus-based affordability ratio, which extends the standard measure by including the electricity expenditures associated with accessing water, primarily for pumping and storage in response to intermittent supply, in addition to direct water charges. Unlike the standard affordability ratio, which divides only the monthly water bill by household income, the nexus-based ratio captures the combined financial burden of water and water-related energy costs for accessing water under intermittent supply conditions. To estimate the fraction of electricity expenditures associated with accessing water, we consider the electricity consumed by electric water pumps to determine the monthly electricity use for water. The calculation of electricity use is based on the pump power, the frequency of pump use, and the duration of each pump operation, assuming that the pumps operate at maximum capacity and accounting only for the use of the primary electric pump in each household. These simplifying assumptions are employed because households may use electric pumps of varying brands, ages, and conditions, making precise electricity consumption data associated with water access difficult to obtain. However, they may slightly underestimate actual electricity use by ignoring pump aging, part-load operation, and any secondary pumps. Consequently, these conservative estimates represent a lower bound; if household electricity expenditure for water access remains substantial even under these assumptions, it further underscores the importance of accounting for these hidden charges. Within the demand function subsample, there are seven missing values for pump power, where households did not provide this information during the survey. We replace these missing values with the median value for electric water pump power from the full sample of standard urban households in Pimpri-Chinchwad. Since an IBT also applies to domestic electricity consumption in the case study area, we assume that electricity used for addressing water supply issues is essential and, therefore, charged at the lowest tariff block.

2.3.2. Demand Estimation

To address the second research question and establish a basis for the third research question, we estimate a household piped water demand function that includes electricity price as an explanatory variable. The region’s highly intermittent municipal piped water supply, as discussed in Section 2.1, increases the demand for household activities that process temporally accessible water into water that can be used at any time, e.g., through the use of rooftop storage tanks. This can lead to a predominant complementary relationship between water and energy, as they are complements in water processing activities [15]. Based on this context, we hypothesize that piped water and electricity are economic complements for Pimpri-Chinchwad households, and thus expect to observe a negative cross-price elasticity with respect to electricity price in our demand estimation.

To operationalize the electricity price variable ($p_e$), we use billing information on piped water and electricity from the 2019/20 financial year in the survey dataset. As briefly introduced in Section 2.2, the billing data for piped water and electricity are provided in different formats. Piped water bills are available for a single period for each household, with billing intervals ranging from two to three months for most observations and over one year for some. By referencing the billing dates, we assign each household’s water consumption to a specific season (i.e., winter, summer, or monsoon) or as an annual average. In contrast, electricity bills include consumption data for multiple periods, covering one to twelve months. We calculate the average monthly electricity consumption for each household by season and for the year as a whole. This allows us to match each household’s piped water consumption with the corresponding seasonal or annual average monthly electricity consumption. Using the 2019/20 electricity tariff structure, we determine the electricity consumption block and the marginal electricity price paid by each household for the period in which their water consumption data falls.

In addition to the electricity price ($p_e$), we include water price ($p_w$), household net disposable income ($y$), and household size ($h$) as explanatory variables. These factors are well-established determinants of residential water demand [21]. We use the marginal price of water to represent the price households respond to in their consumption decisions. However, under IBT, households face a higher marginal price as their consumption moves into higher blocks, while the average price they pay for water is subsidized due to lower rates for initial consumption. To accurately capture the effect of income, we introduce a difference correction ($d$) within the income variable, which reflects the difference between the actual water bill and the hypothetical bill if all units were charged at the marginal rate [44,45,46,47]. For household composition, we include both household size ($h$) and its square ($h^2$) to account for potential scale effects.

Regarding the functional form, we considered several functional forms for the demand estimation, including log-linear, double-log, and Stone–Geary models. The Stone–Geary demand function offers econometric advantages, as it separates consumption into a fixed subsistence quantity that is unaffected by price and a variable component that responds to price changes [48,49,50]. This approach is theoretically appropriate for essential goods such as water, where a minimum level of consumption is necessary to satisfy basic needs. However, in Pimpri-Chinchwad, as introduced in Section 2.1, households receive the first six cubic meters of water per month free of charge. With an average household size of about five members (see Table 2), this is equivalent to free water of 40 LPCD, matching the critical water-use threshold for vulnerability [8,51]. The zero price for the first block of consumption ensures that subsistence needs are already met, which complicates the application of the Stone–Geary form in this context. This zero-price feature also presents challenges for the double-log form, which has the advantage of explicitly reporting both income and price elasticities [21,22,52]. The double-log functional form, as the name suggests, is only appropriate when all prices are positive, and is therefore not feasible in this case. Regarding the log-linear form, although it does not yield a constant price elasticity, studies have shown that it provides unbiased estimates compared to the double-log form [53,54]. Therefore, we employ a log-linear functional form for our demand estimation, as specified in Equation (2) below, where ${\beta }_i$ denote coefficients for the variables and $u$ represents the error term:

$$\ln q={\beta }_0+{\beta }_1{p}_w+{\beta }_2\ln p_e+{\beta }_{3}ln\left(y+d\right)+{\beta }_{4}h+{\beta }_5{h}^2+u$$

(2)

For the dependent variable, we use the metered consumption data from water bills provided in the survey dataset to determine the amount of piped water demanded by each household. As previously explained, piped water in the case study region was billed at irregular intervals, varying from two months to over a year. Therefore, we convert the billing data to daily consumption. The dependent variable is defined as the natural logarithm of the quantity of daily per capita piped water consumption within a household ($q$).

The explanatory variables are as follows: (i) the marginal price of piped water $p_w$, whose coefficient ${\beta }_1$, together with the tariff structure, yields the own-price elasticity of demand for a given price (see Equation (3) below); (ii) the natural logarithm of the marginal price of electricity for the household $p_e$, with its coefficient ${\beta }_2$ representing the cross-price elasticity between electricity and piped water; (iii) the natural logarithm of monthly household income $y$ corrected with the difference variable $d$, with its coefficient ${\beta }_3$ indicating the income elasticity; (iv) household size $h$ and (v) its square term $h^2$.

$$e_{q,p_w}={\displaystyle \frac{\partial q}{\partial p_w}}{\displaystyle \frac{p_w}{q}}={\beta }_1{p}_w$$

(3)

We employ a DCC model to estimate the demand function. Under IBT pricing, households face progressively higher marginal prices as consumption increases, indicating a correlation between price and consumption levels. This interdependence introduces endogeneity bias, which makes standard ordinary least squares (OLS) regressions unreliable, e.g., it can produce counterintuitive positive price elasticity estimates. The DCC model resolves this issue by simultaneously estimating household decisions regarding tariff block selection (i.e., discrete choice) and consumption quantity (i.e., continuous choice). Compared to instrumental variable methods, which have also been widely used to address endogeneity issues, the DCC model has the advantage of explicitly modeling price changes at block boundaries (i.e., the kink points) and providing more accurate price elasticity estimates when household preferences strongly influence demand patterns [53,55,56,57]. The DCC model uses the maximum likelihood estimation approach to estimate the demand function with two error terms ($\eta$ and $\epsilon$), and modifies Equation (2) as follows:

$$\ln q={\beta }_0+{\beta }_1{p}_w+{\beta }_2\ln p_e+{\beta }_{3}ln\left(y+d\right)+{\beta }_{4}h+{\beta }_5{h}^2+\eta +\epsilon$$

(4)

In Equation (2), the single error term $u$ captures unexplained variation in demand, i.e., the factors affecting consumption that are not measured or included in the model. The DCC specification in Equation (4), however, separates this disturbance into two components: the heterogeneity error η, which reflects unobserved household preferences that the consuming households themselves understand but researchers cannot fully observe or quantify in the estimation; and the perception error ε, which represents the unobservable factors to both the consuming households and the researchers that influence the final observed consumption level [58]. The heterogeneity error drives the discrete decision of whether to consume water at all or which tariff block to select, while both error terms affect the continuous quantity consumed once participation is decided [58]. By distinguishing these two sources of randomness, the DCC model more accurately reflects household water decisions under block-rate pricing: first choosing whether and at which tariff block to consume, and then determining the consumption quantity.

In addition to the DCC model, we construct two OLS models for comparative validation, following the methodology that Klassert et al. [53] applied to estimate household water demand in Jordan. The OLS models use the demand function expression in Equation (2); one includes the water price variable ($p_w$), while the other excludes it. It should be noted that the OLS model with the water price variable is subject to bias due to endogeneity, while the model excluding the water price variable can only provide limited insight into the effects of non-water-price factors, as it does not account for the price impact. Although neither OLS model yields reliable standalone estimates, they provide useful contrasting points that help validate the direction and magnitude of the non-water-price coefficients estimated by the DCC model.

2.3.3. Microsimulation of Price Impact

Further addressing the third research question on the impacts of water and electricity price changes, to exemplify how demand estimates can inform pricing policies, we construct a microsimulation model based on the coefficients estimated from the DCC model to project household water demand under hypothetical price increases. This illustrative example assumes consumption blocks remain the same and focuses on rate adjustments. As outlined in Section 2.1, although water prices in Pimpri-Chinchwad have increased modestly in recent years, future demand management may require more aggressive pricing strategies, in particular, to address households with excessive consumption. To explore these possibilities, we test two water price scenarios: (1) price increases across all consumption blocks and (2) price increases only to the upper three consumption blocks (i.e., to control monthly household consumption exceeding 15 m³). In the scenario where water price increases only for the upper consumption blocks, we also test how electricity price increases may affect household piped water demand.

The simulation employs the same subsample of 160 households that is used for demand estimation. For each household, we incorporate the observed error terms from the estimation process and adjust the consumption blocks iteratively to align with the projected demand quantities. We then evaluate two key outcomes for each pricing scenario: (1) the average household water consumption in LPCD and (2) the proportion of households falling below the critical vulnerability threshold of 40 LPCD.

3. Results

3.1. Water Affordability in Pimpri-Chinchwad

Regarding our first research question, integrating the nexus perspective reveals greater potential affordability issues and provides deeper equity-based insights into water affordability assessments in areas with intermittent water supplies. As shown in Table 2, the average duration of daily piped water supply in Pimpri-Chinchwad is 3.63 h, with a wide range from just 0.5 h up to 24 h per day. This highlights the highly intermittent nature of the piped water supply in the area. To cope with this intermittency, electric pumps are widely used. Among the 568 standard urban households surveyed, 51.9% reported using an electric water pump. On average, these pumps are operated about 1.08 times per day, running for approximately 67.4 min each time.

We calculate electricity costs for accessing water with the demand estimation subsample based on pump power, frequency, and duration of pump operation. The average monthly water bill for these households is 116.2 INR; however, when electricity costs for accessing water are considered, the average total water-related expenditure rises to 218.2 INR per month, almost double the water bill alone. Notably, households that fall in the lowest consumption block (i.e., those using less than 6 m³ of piped water per month) do not need to pay in terms of water bill, yet they still incur electricity expenses for accessing water. Overall, 36.9% of households spend more on electricity for accessing water than on the water bill itself.

Despite these additional costs, our calculations of the conventional affordability ratios do not indicate a widespread water affordability issue in Pimpri-Chinchwad, likely due to the region’s relatively low prices for water and electricity. Only 1.25% of households exceed the 2.5% affordability ratio threshold when considering the amount of water bills alone; this figure rises to 3.13% when electricity costs are included in assessing the nexus-based water affordability ratio. To test the robustness of these results, we conduct sensitivity analyses using thresholds between 2% and 5%, as recommended in water affordability literature [39,40,42,43]. The share of households exceeding these thresholds ranges from 0% to 1.25% when considering water bills alone, and from 0% to 3.13% when including electricity costs for accessing water. At the 5% threshold, no households in the analyzed sample are observed to have affordability issues using either approach. At the 3% threshold, no households face affordability issues when accounting for the standard affordability ratio, while 1.88% are still considered to have unaffordable water when accounting for the nexus-based affordability ratio. Although the differences observed between the standard and nexus-based affordability ratios appear modest in absolute terms, they represent a considerable number of people, given the city’s large and growing population. Taking the 2011 Census of India estimates [25], for example, the 1.88% difference could refer to more than eight thousand households affected. Moreover, as discussed in Section 2.3.1, our calculations assume pumps run at maximum capacity and only account for the use of the primary pump, even though households may own multiple pumps. In reality, the actual electricity usage and associated costs for accessing water could be higher, potentially making water unaffordable for even more urban households.

Comparing household water expenditures across income levels reveals further insights into equity issues. Figure 1 shows the average household water expenditures according to water bills and expenditures including electricity costs for accessing water by income quintile. While differences in the amount of water bills among the bottom 80% of households are minor, the top quintile stands out due to its much higher water consumption and, accordingly, the expenditures according to water bills. The wealthiest 20% households, who have greater purchasing power and higher consumption levels, have an average monthly water bill of 216.7 INR. This is nearly double that of the second-wealthiest quintile and 2.5 times as high as that of the poorest quintile. However, when electricity costs for accessing water are considered, the wealthiest households do not bear proportionately increased water expenditures despite their high consumption. This is likely because they benefit from a more reliable water supply and therefore have less need for coping strategies: as noted by Zhu, Gawel et al. [9], wealthier households in the urban areas of the Pune Metropolitan Region generally receive longer daily hours of piped water supply compared to lower-income households. By contrast, the lowest-income households also do not spend much on electricity for accessing water; they have the lowest electricity expenditures for water. Different from the top quintile, this is probably due to the lack of capacity to invest in energy-intensive coping mechanisms. Instead, these households may rely on time-consuming strategies, such as waiting at taps to fill water containers manually or collecting water from alternative sources, as discussed in conceptual nexus analyses [15] and demonstrated in empirical case studies in India [14,59], Nepal [17], and Kenya [16]. The largest gap between water bills and total water-related expenditures is found among middle-income households (i.e., quintiles 2–4). These households receive a highly intermittent water supply and have the resources to invest in energy-intensive coping mechanisms, such as electric pumps with rooftop storage. Their electricity expenditure for accessing water can exceed their water bill by more than double. It is doubtful that these households are fully aware of the substantial electricity costs for accessing water. While this may not reach the threshold of a critical affordability issue for middle-income households, it represents a significant financial burden for them that warrants more attention.

3.2. Estimates for Household Piped Water Demand

With regard to our second research question on the economic relationship between electricity and water, to analyze the factors impacting household piped water demand, we estimate a DCC model and two OLS models for comparison.

Table 3. Estimates of the Natural Logarithm of Household Per Capita Daily Piped Water Demand in Pimpri-Chinchwad.

Dependent Variable
	Log (LPCD)
	OLS		DCC
	(1)	(2)	(3)
Marginal Water Price	0.120 ***		−0.070 ***
	(0.007)		(1.722 × 10−3)
Log (Marginal Electricity Price)	0.114	−0.092	−0.097 ***
	(0.073)	(0.127)	(4.888 × 10−3)
Log (Income + Difference)	0.043	0.166 ***	0.177 ***
	(0.036)	(0.061)	(4.830 × 10−4)
Household Size	−0.326 ***	−0.252 ***	−0.247 ***
	(0.035)	(0.060)	(6.010 × 10−4)
Square of Household Size	0.011 ***	0.009 **	0.007 ***
	(0.002)	(0.004)	(2.259 × 10−4)
Constant	4.828 ***	4.500 ***	4.877 ***
	(0.360)	(0.630)	(7.993 × 10−3)
Observations	160	160	160
R2	0.755	0.245
Adjusted R2	0.747	0.225
Residual Std. Error	0.293 (df = 154)	0.512 (df = 155)
F Statistic	94.835 *** (df = 5; 154)	12.544 *** (df = 4; 155)
SigmaEta			−6.781 × 10−3 **
			(3.156 × 10−3)
SigmaEps			−4.087 × 10−1 ***
			(1.891 × 10−2)
Log-Likelihood			−1.591 (df = 8)

Note: * p < 0.1; ** p < 0.05; *** p < 0.01.

presents the estimation results of these models. As discussed in Section 2.3.2, the two OLS models only serve as reference points for the non-water-price variables; their coefficients are unreliable for further analysis. The OLS (1) model includes the observed marginal water price, but its coefficients are biased due to endogeneity introduced by the IBT structure. The OLS (2) model is a baseline model that excludes the water price variable and partially captures the effects of non-water-price factors on piped water demand. In contrast, the DCC model addresses endogeneity and provides robust estimates for the demand function. Table 3 shows that all coefficients estimated by the DCC model are statistically significant, and the effects of the non-water-price variables are consistent with the estimates of the OLS models in terms of signs and magnitudes.

The DCC model shows that household demand for piped water is income inelastic, with an estimated income elasticity of 0.177. This means that as household income increases, per capita water consumption rises, but at a rate that is less than proportional. This is also reflected previously in Figure 1 in terms of the water bills. Household size also has a significant impact on water demand: as the number of household members increases, per capita water use declines. While the total household water use may increase with more members, the per capita water use decreases due to scale effects, e.g., with shared water activities and efficiencies. Specifically, per capita daily water consumption decreases by 24.7% with each additional household member, holding other factors constant. However, the magnitude of the scale effect diminishes in larger households, as indicated by the small but positive coefficient on the squared term of household size. This suggests that, for very large households, the reduction in per capita water demand slows down, possibly due to the saturation of shared activities or increased discretionary consumption in extra-large households.

The DCC model’s estimates also identify significant impacts from water prices. Since the demand function is in log-linear form, the price elasticity is not a constant and varies across tariff blocks.

Table 4. Water Price Elasticities by Tariff Block in Pimpri-Chinchwad based on the Price Coefficient from the DCC Model.

Tariff Block	Price Elasticity
Block 1 (0 INR/m3)	0
Block 2 (4.2 INR/m3)	−0.292
Block 3 (8.4 INR/m3)	−0.584
Block 4 (12.6 INR/m3)	−0.876

summarizes the estimated price elasticities for each block. Here we combine consumption blocks 6–15 m³ and 15–22.5 m³ into one block because, although the Pimpri-Chinchwad Municipal Corporation officially sets five consumption blocks for residential water users, the prices for these two blocks are always the same. Therefore, they would not impact households’ decisions to switch consumption levels. This combined consumption block is also where the majority of households’ consumption falls. The price elasticity of the block is estimated at −0.292, indicating households are limitedly responsive to price changes. While this value is slightly lower in absolute terms than the empirical findings in other Indian cities, i.e., −0.464 in Chennai [60] and −0.435 in Jaipur [61], it remains close to the −0.3 to −0.6 range of price elasticities for developing countries identified by Nauges and Whittington [22]. The results on water price elasticities indicate that demand is inelastic in lower consumption blocks and becomes more elastic in higher blocks. This suggests that households consuming less water have limited flexibility to reduce their use in response to price increases, while those in higher consumption blocks are more sensitive to price changes.

In addition, the DCC model estimates a significant cross-price elasticity of −0.097 between electricity and piped water. The negative sign of the coefficient indicates a complementary relationship: as the electricity price rises, water demand decreases. However, the magnitude of the coefficient suggests that this impact is limited. For instance, doubling the electricity price would reduce average household piped water demand by less than 10%. Given the current marginal electricity price for the lowest block, an increase of 1 INR would lead to an estimated 1.78% decrease in household piped water demand. For an average household, this equates to a reduction of approximately 2.91 LPCD. This weak complementarity brings us back to the concern raised in Section 3.1 that households may not fully recognize the considerable electricity costs associated with accessing water beyond their water bills. On the other hand, as indicated by Zhu, Gawel et al. [9], in the urban areas of the Pune Metropolitan Region, electricity used for accessing water can account for on average up to 27% of total household electricity consumption. It is also highlighted in Section 3.1 that the electricity costs associated with water intermittency can be as much as the water bill itself, doubling total water-related expenditures for a large share of households. This is an issue that still needs to be addressed in the formulation of water demand management strategies. While electricity prices influence water demand due to the need to cope with an intermittent supply, they are not a strong lever for demand management. Unless water supply infrastructure improves, electricity pricing alone is unlikely to substantially affect residential household water use. This further emphasizes the need for thorough consideration of water pricing strategies that take into account the inseparable nexus perspective.

3.3. Impact of Price Increases on Household Piped Water Demand

Findings from Section 3.1 and Section 3.2 reveal that wealthier households consume more water and that those in higher consumption blocks are more sensitive to price changes. Based on these findings and to address our third research question on the impact of price changes, we simulate household piped water demand under hypothetical price increase scenarios to exemplify how results from the DCC model can inform demand management strategies with equity concerns. Figure 2 presents the results of two hypothetical water price increase scenarios. As detailed in Section 2.3.3, the scenario with price increases only for upper consumption blocks targets households consuming more than 15 m³ per month.

As shown in Figure 2, in both scenarios, increasing water prices leads to reductions in household water demand and higher proportions of households falling below the critical vulnerability threshold. The scenario where all blocks see price increases represents an extreme case: as prices rise, average household water consumption could approach zero, causing all households to become vulnerable. When prices increase by less than three times, both scenarios have a similar effect on regulating water demand. For instance, doubling all water prices would reduce average household water demand from 184.10 to 107.86 LPCD. Meanwhile, doubling prices only for the upper consumption blocks would reduce the average demand to 109.44 LPCD-a very similar outcome. At this stage, both scenarios are effective in curbing excessive water use without significantly increasing water vulnerability. In both scenarios, doubling prices would result in about 4% of households consuming less than 40 LPCD, compared to the current rate of 1.88%. However, if prices increase threefold or more for all blocks, water demand drops sharply, and the proportion of vulnerable households increases much more rapidly. In contrast, tripling prices or more for only the upper blocks results in a more gradual decline in demand, with relatively higher average consumption and lower vulnerability rates. Although these scenarios are hypothetical extremes, the results suggest that targeting price increases at high-consumption households offers better protection for lower-consumption households against the negative effects of price hikes.

We also examine the combined effect of water and electricity price increases based on the scenario where only the upper block water prices rise, as illustrated in Figure 3. The results show that higher electricity prices further reduce water demand, but the effect is limited. For example, when water prices for the high consumption blocks are doubled, doubling the electricity prices leads to less than a 5% reduction in average water demand, while a tenfold increase in electricity prices results in a roughly 13% reduction. The impact on the share of households below the vulnerability threshold is relatively small at lower water price multipliers, but becomes more pronounced when water prices are increased more aggressively.

The results from Figure 2 and Figure 3 show that, as more households fall below the 40 LPCD threshold at higher price multipliers, concerns about water security and equity become more urgent. This is especially true if price increases are not paired with targeted support for vulnerable groups. Effective water demand management must balance conservation goals with social protection [30,31,32]. Although broad price increases can achieve greater conservation, they risk harming vulnerable groups, as water expenditures are associated with broader issues of insecurity, such as food and health [62]. Targeted increases for higher consumption levels moderate these impacts but result in less overall demand reduction. From an equity perspective, raising prices only for the upper consumption blocks is preferable, as it regulates demand without disproportionately affecting low-consumption, often lower-income, households. Restructuring consumption blocks with targeted controls on specific consumption ranges could help control overconsumption while safeguarding water use for lower consumption groups, e.g., imposing larger increases for the highest block and moderate increases for middle blocks. The combined effects of water and electricity pricing on household welfare should also be considered in water demand management. Raising water prices reduces demand and increases vulnerability; simultaneous electricity price hikes amplify these effects, especially when water prices rise sharply. The combined burden of higher water and electricity prices may push a significant share of households below critical water use levels. These effects must be considered when determining water and electricity pricing strategies.

4. Discussion

4.1. Hidden Costs and Water Affordability

In relation to our first research question on the influence of adopting a water–energy nexus perspective on affordability assessments, our results suggest that the true financial burden of accessing water under intermittent supplies extends beyond water bills. Incorporating this nexus perspective can therefore enhance water affordability assessments by providing equity-based insights. Our household-level analysis empirically supports previous conceptual discussions that measurements of water affordability should consider all costs for households to obtain safe water and adequately address equity goals [38,63].

In Pimpri-Chinchwad, many households face substantial, and often unrecognized, electricity costs associated with accessing water. Notably, approximately one-third of households spend more on electricity to access water than on their water bills alone, revealing hidden affordability challenges. These additional costs disproportionately affect middle-income households. Unlike wealthier households, which benefit from a more reliable water supply, middle-income households often need to cope with frequent supply interruptions. At the same time, they have the resources to invest in energy-intensive coping equipment, such as large storage and electric pumps, in contrast to lower-income households, which often lack the capital to do so. This results in higher electricity costs for water access. However, lower electricity costs do not necessarily mean a lighter burden for the poor households. Instead, these households may rely on time-consuming manual collection methods or reduce their overall water use, as discussed in both conceptual and empirical analyses [15,16,17,59]. It is important to note that our analysis focuses on standard urban households and does not capture the situations experienced by extremely poor households living in informal settlements. Conditions are likely more severe in informal settlements, where water access conditions are typically worse than observed in the study sample. Research on water use in informal settlements in Pune, for example, documents significant water insecurity and considerable time spent collecting water, with some households unable to satisfy basic needs [14]. In addition, our calculation of electricity expenditures assumes that pumps operate at maximum capacity and accounts only for the primary pump. In practice, aging pumps often operate below full design capacity, and some households may use multiple pumps. Therefore, our estimates likely understate actual electricity consumption and costs, implying that the financial burden of intermittent supply could be even greater and affect more households than our analysis captures. Nonetheless, this study provides new empirical evidence of the magnitude of these hidden costs and highlights the distinct distributional impacts of coping with intermittent supplies, as these burdens differ substantially across income groups.

Despite the significant additional costs associated with water, widespread water affordability issues are not currently evident in Pimpri-Chinchwad, likely due to relatively low water and electricity prices. When calculating affordability ratios, using the nexus-based water affordability ratio that takes electricity expenditures for accessing water into account results in a modest increase in the proportion of households facing affordability issues. Although this percentage increase is limited in absolute terms, it can translate to a large number of people, given the city’s population size and rapid growth. It is important to note that our affordability analysis is based solely on the conventional affordability ratio approach due to data limitations. While this method is favored for its straightforward application and moderate data requirements [38,40,41], it has known limitations, e.g., it does not distinguish affordability from poverty since a low ratio may mask deprivation if households reduce water use below essential needs to remain within an “affordable” threshold; it also fails to capture non-monetary costs such as the time and effort involved in accessing water, and these factors are especially relevant in settings with intermittent supply conditions [37]. These methodological limitations underscore the need for future research to employ and combine a broader set of affordability assessment approaches that better reflect varying service conditions and distributional effects across different household groups. However, by introducing a nexus-based variant of the conventional affordability ratio, this study provides empirical insights into the benefits of explicitly including the demand-side water–energy nexus perspective in household affordability assessments. Overall, the hidden costs and affordability challenges revealed here underscore the necessity of a comprehensive nexus-aware assessment of water affordability that considers both direct and indirect expenses, particularly in regions with intermittent water supply systems.

4.2. The Water–Energy Nexus and Tariff Design

Our econometric analysis indicates that household piped water demand in Pimpri-Chinchwad is generally inelastic with respect to income and water price at typical consumption levels. Specifically, the income elasticity is estimated at 0.177, while water price elasticities range from 0 to −0.292 for consumption below 150 LPCD. However, the piped water demand becomes more price-responsive at higher consumption levels, where water price elasticities range between −0.584 and −0.876. These findings contribute empirical evidence illustrating household piped water demand patterns in developing countries, particularly in India, where such analyses are rarely available due to data constraints [22,60,61,64]. Our study extends the literature by providing reliable estimates of income and price elasticities, as well as household size effects from a DCC model, which can serve as references for modeling integrated urban water–energy systems, e.g., with agent-based models, and for comparative research across diverse regional contexts.

Addressing our second research question on the cross-price elasticity between electricity and piped water, the DCC model’s estimates confirm a complementary economic relationship between electricity and piped water, driven by the extensive use of electric pumps to cope with the intermittent supply in Pimpri-Chinchwad. This complements conceptual analyses of the demand-side water–energy nexus at the household level in the literature [15]. The estimated cross-price elasticity coefficient of −0.097 suggests that a 1% increase in electricity price results in less than 0.1% decrease in piped water demand. While this indicates a weak complementarity, the relatively small effect implies that electricity pricing alone is unlikely to be an effective tool for managing water demand, especially without improvements in water supply infrastructure. Notably, previous estimates of cross-price elasticities between water and energy have been recorded in developed countries, e.g., by Hansen [65] for Copenhagen, Maas et al. [66] for Colorado, and Suárez-Varela [67] for Spain. To our knowledge, such estimates have been rarely reported for developing regions. Our study fills an important empirical gap and adds to the quantitative understanding of the economic relationship between water and energy in regions with an intermittent water supply.

Simulations of hypothetical pricing scenarios further exemplify the insights from the DCC model’s estimates and illustrate the impact of water and electricity prices on piped water demand, addressing our third research question. The results demonstrate that broad-based water price increases can reduce demand, but they also risk pushing more households below the critical water use threshold. In contrast, targeted price increases for higher consumption blocks can achieve similar reductions in excessive use due to the higher price elasticity observed in these blocks, while better protecting lower-consumption, often lower-income, households. Combining water price increases with electricity price increases exacerbates the impact on household vulnerability, especially at higher water price levels. These findings align with literature highlighting the benefits of IBT and integrated tariff designs for balancing conservation objectives with social equity, while also emphasizing the need for caution to address potential unintended consequences related to targeting and fairness [68,69,70,71,72,73,74]. Furthermore, our study contributes to the empirical understanding of the direct impact of electricity pricing on water demand. Although this impact is limited, it should not be overlooked, especially in conditions of extreme price fluctuations or crises that exacerbate water supply insecurity.

As demonstrated in previous empirical studies [9,16,17] and supported by our affordability assessments, substantial electricity consumption is associated with accessing water under intermittent supply conditions. However, the weak complementarity identified by our econometric model suggests that the impact of the electricity price on piped water demand is rather limited. It is possible that households may not fully account for these associated costs in their water-related decisions. Without improvements to water supply infrastructure, it is unlikely that the substantial electricity consumption for dealing with water intermittency will change. Therefore, municipal authorities should recognize that intermittent supply systems impose a significant burden on households, a challenge that has not received much attention so far. Transparent information about these combined water and electricity costs is essential for households to avoid underestimating their total expenditures, as well as for policymakers to understand the full extent of water insecurity and affordability challenges. Addressing these challenges requires a comprehensive urban water management approach that incorporates the water–energy nexus and ensures both water and energy services remain accessible and affordable. In particular, water management strategies and decisions on pricing and improving water supply infrastructure should be developed in a more sophisticated and integrated way, thoroughly considering the nexus perspective.

5. Conclusions

This study advances the empirical understanding of the household-level demand-side water–energy nexus under intermittent piped water supply conditions with a case study of Pimpri-Chinchwad, India. Our analysis reveals that many households incur substantial and often hidden electricity costs to cope with unreliable water supply; for approximately 37% of households, these electricity expenses exceed their direct water bills. Middle-income households are particularly burdened, as they both face intermittent supply and have the means to invest in energy-intensive coping mechanisms. Conversely, lower-income households may lack such options, potentially leading to extensive time costs for collecting water or inadequate water use. Although widespread affordability issues are not currently evident in Pimpri-Chinchwad, mainly due to low resource prices, the results imply that a large share of the population could be at risk when energy prices rise or supply conditions deteriorate. As this study focuses on standard urban households due to data limitations, greater affordability issues may exist in the region than observed here, particularly in informal settlements.

Our econometric analysis reveals significant impacts of water and electricity prices, income, and household size on household demand for piped water. Household piped water demand is inelastic to income and water price at typical consumption levels, but becomes more price-responsive at higher consumption blocks. We identify a weak complementarity between electricity and piped water, suggesting that households’ expenditure on electricity for water access is a rational response to supply intermittency. However, the relatively small magnitude of the impact of electricity prices on piped water demand implies that households may not fully account for electricity costs when making water-related decisions despite their substantial expenditures. The estimated price elasticities, as exemplified by the simulation results, indicate that tariff structure designs should be progressive and targeted, focusing price increases on higher consumption blocks to curb excessive water use while protecting vulnerable groups. Combined increases in water and electricity prices can exacerbate vulnerability, particularly in the absence of targeted support measures or improvements in water supply infrastructures.

Overall, our findings empirically confirm the relevance of the demand-side water–energy nexus under intermittent piped water supply conditions in Pimpri-Chinchwad. These results highlight the importance of integrating nexus considerations into water affordability assessments and urban water demand management strategies. Policymakers and utilities should recognize that the financial burden imposed on households by intermittent supply systems may extend beyond direct water charges and can include substantial, though frequently unaccounted for, electricity expenditures. Such costs can be underestimated by households themselves and are rarely captured in conventional affordability metrics. Future urban water management, including planning for tariff reform, infrastructure development, and policy instruments, could benefit from accounting for these hidden costs. For instance, when evaluating the affordability of new pricing structures across income groups, it may be helpful to include infrastructure-related coping costs, such as electricity used for pumping and storage, in the assessments, particularly in areas with frequent interruptions or low water pressure. Moreover, identifying households or neighborhoods facing poor water service reliability and accordingly high coping expenses could enable utilities to implement targeted subsidies or billing rebates, thereby preventing pricing reforms from exacerbating inequalities. This nexus perspective may also inform municipal decisions on infrastructure upgrade priorities. Investments in expanding municipal storage capacity and improving supply schedule reliability that reduce the frequency and severity of supply interruptions may not only improve water services but also increase overall system efficiency by reducing unnecessary electricity consumption at the household level. Aligning long-term infrastructure planning with cost recovery and affordability goals could contribute to more sustainable and equitable urban water systems.