Rural drinking water supply program and societal development: Evidence from the early implementation phase of India’s Jal Jeevan Mission

1.1. Implementation phases of Jal Jeevan Mission

Jal Jeevan Mission is currently under implementation in a time-bound mission mode to ensure functional household tap connection at each rural household with adequate quantity (55 litres per capita per day) and prescribed quality (BIS:10500) on regular basis by 2024. Although JJM is a nationwide program, the states and union territories are given autonomy to implement the program based on their State Action Plan (SAP) with timelines for a 100% coverage [4]. The program launched in August 2019, started its implementation process in early 2020 following a batchwise implementation of schemes covering villages in each district. Broadly, the program aims to create water supply infrastructure in two categories: i) creation of in-village infrastructure or single village scheme (includes source development/ augmentation/ grey water management) and ii) infrastructure for bulk transfer of water or multi village scheme (includes treatment and distribution system). Creation of infrastructure for various schemes under the program may take 12 to 18 months and can be implemented in following three phases:

Planning and mobilization phase

The planning and mobilization phases start from identification of village by DWSM (District Water and Sanitation Mission) and last till the approval of estimate for implementation of in-village infrastructure by DWSM taking approximately 3–6 months of time. This phase includes activities such as baseline mapping of existing tap connections, identification of villages, assigning department official for the implementation, community familiarization with the JJM objective, constitution of sub-committee under gram panchayat (Village Water and Sanitation Committee (VWSC) or Pani Samiti), preparation and approval of VAP (Village Action Plan) and planning for work execution.

Implementation phase

The implementation phase takes about 6–12 months of time and includes the ground level initiation of construction based on the scheme identified. The major activities in this phase include testing of quality, augmentation of water source, creation of water supply infrastructure till household tap connections, greywater management, geo-tagging of assets, creation and maintenance of register for accounts, trial runs, installing water measurement devices (meter sensors), fixing and collection of community contribution/ water tariff.

Post implementation phase

The post implementation phase commences after the development of water supply infrastructure and last for 3–4 months. Major activities of this phase include water supply to households, operation and management, water tariff collection and capacity building.

Considering the time taken to develop infrastructure following various phases of implementation, 13.39% of villages were identified as ‘Har Ghar Jal (HGJ) reported’ (villages which are reported to be operational and yet to be approved by Gram Panchayat) in 2020, and this number escalated to 22.15% in 2021 [1]. However, there was no ‘HGJ certified’ villages (fully functional and approved by Gram Panchayat) during our study timeline. The first group of villages declared ‘HGJ certified’ were in 2022 which lies beyond the data collection timeline (2019–21) in our study. Hence, we anticipate the expected effect in our study to be the representation of ‘HGJ reported’ villages during 2019 and 2021. Moreover, since JJM is large program implemented across nation in a phase wise manner, the actual effect (especially in long-term) is expected to be realized only after completion of the project. Hence, the data period considered in our study aims to reflect the early/intermediate effects of JJM in terms of access to drinking water as shown in Fig 2. A detailed illustration of implementation guideline of JJM comprising of strategy for planning and implementation, institutional mechanism, implementation phases, financial planning/funding, technological interventions/innovations, support activities, water quality monitoring and surveillance, and monitoring/evaluation can be found elsewhere in the operational guidelines for the implementation of JJM [4].

2.1. Study design and setting

This study is a before and after cross-sectional analysis based on data from two different time periods: 2015–16 and 2019–21. We included all the Indian states in our analysis except the union territories due to the extremely low number of observations. Further, we categorized states into i) High focus Empowered Action Group States (EAG), ii) Low focus states (other than EAG and NE states), and iii) North-eastern (NE) states.

2.2. Data and variables

We utilized nationally representative sample survey data from round 4 (2015–16) and round 5 (2019–21) of the National Family Health Survey (NFHS). NFHS data from various rounds was accessed from the 7^th phase of Demographic Health Surveys (DHS). The surveys were conducted under the stewardship of the Ministry of Health and Family Welfare (MoHFW) [12] and the designated nodal agency- International Institute for Population Science (IIPS) [13]. The survey identifies a sample of villages (created considering the village population and the percentage of the population belonging to scheduled castes and scheduled tribes) as primary sampling units (PSUs) and follows a two-stage sampling method, which includes selection of PSUs using probability proportional to size (PPS) sampling in the first stage and random selection of households using systematic sampling in every selected cluster in the second stage. More detailed information on data collection techniques can be accessed from IIPS’s portal [14–17].

We considered household data sets from various rounds of NFHS and included only rural households in our study. In the fourth round of NFHS, data were collected from 601,509 households. Of these, 179,844 households belonging to urban areas and UTs were excluded, and a remainder of 421,665 households were selected as the sample in our study. The fifth round of NFHS collected data from 636,699 households, of which we excluded 164,593 households belonging to urban areas and UTs. Further, to capture the effect of JJM, we dropped 81,892 households interviewed prior to the launch of JJM (August 2019). This resulted in a final sample of 390214 households.

We selected four variables covering different aspects of access to drinking water: i) households with tap water connections in the premises, ii) adult women (age 15 to 49) in the household fetching drinking water from distant sources, iii) female children (up to age 15) in the household fetching drinking water from distant sources and iv) time elapsed in fetching water from distant sources. To estimate the program effect, the variable ‘households with tap water connections in the premises’ was used as a proxy for JJM intervention. In order to generate the above variables, we explored several drinking water-related questions that were asked to the households, such as: What is the main source of drinking water for members of your household?; Where is the water source located?; Who usually goes to this source to fetch water for your household?; How long does it take to go there, get water, and come back in one trip?.

2.3. Analytical tools

We employed multiple analytical tools to assess the inter-state variation, determinants, and district-level effect of access to tap water connection. We started with the calculation of state-level proportions of each of our components covering each state and state group and presented the estimates using graphs and descriptive statistics. Thereafter, we estimated a set of probit models to identify the determining socioeconomic and demographic factors in the access to drinking water. The outcome is represented by a latent variable $\left(Y_h^{*}\right)$, explained by a vector of explanatory variables (X_h) such as wealth quintile, highest education level in the family, caste, religion, age of the family head, household size, and state-level fixed effect, which is expressed in the following form:

Here, $Y_h^{*}$is a latent variable representing all four variables related to accessibility of drinking water. The accessibility of drinking water in rural areas was assessed using demand side as well as supply-induced indicators. The demand side indicators include the proportion of household in which women fetching drinking water from distant sources, proportion of household in which female children fetching drinking water from distant sources, and the proportion of households spending more than 30 minutes to fetch drinking water from a distant source. Meanwhile, the supply-induced indicator of accessibility is the household tap water connection (HTC) which is attributable to JJM.

Subsequently, we employed a multi-stage analysis at the district level using a set of multiple linear regression models, where we treated household tap water connection (HTC) as an exposure variable and remaining accessibility indicators as outcome variables. First, we calculated the district-level weighted average, grouped by socioeconomic status and district, for each of our outcome variables and potential confounders identified from the probit models. Further, we matched the observations of 2015–16 with 2019–21 based on unique district identifiers. Second, the change in outcome and exposure variables were calculated by taking the percentage point absolute difference in the mean. Third, the average district-level effect of JJM was estimated using two multiple linear regression models: a change score model (takes into account change in both outcome and exposure variable) and a static model (takes into account change score of the outcome variable and a static form of exposure variable). Our first model of change score estimate takes the following form:

The second model of static estimate takes the following form:

Δy_d is the change in outcome variables between 2015–16 and 2019–21, ΔHTC_d represents the change in the coverage of household tap connections between 2015–16 and 2019–21, HTC2019_21_d represents the coverage of household tap connections in 2019–21 and X_d is the vector of covariates included in the models such as wealth, education and caste of the households. All the statistical analyses were performed using STATA version 17.

3.1. Interstate heterogeneity in access to drinking water

The proportion of rural households having tap water connections in their premises is presented in Fig 3. At the national level, the proportion of households with tap water connections increased to 21.62% in 2019–21 from 18.33% in 2015–16. Much of this increment was noticed among the north-eastern states (from 32.22% in 2015–16 to 47.72% in 2019–21) followed by high-focus EAG states (12.41% in 2015–16 to 17.11% in 2019–21). This increase in household level tap water connection was evident for most of the individual states except for a few states like Goa, Gujarat, Assam, Nagaland, and Sikkim, where the data indicates that the proportion of rural households with tap water connection declined over time.

Fig 4 shows the proportion of households at national and state level in which drinking water was fetched from a distant source by adult women (between age 15 to 45), female children (up to age 15), and households which took more than 30 minutes to fetch drinking water.

The proportion of adult women fetching drinking water declined at the national level from 82.25% in 2015–16 to 75.89% in 2019–21. This decline was mostly observed among the low focus states (77.77% in 2015–16 to 59.11% in 2019–21), followed by the high focus EAG states (86.13 in 2015–16 to 83.43% in 2019–21). A similar pattern of decline was noticed for the majority of the individual states except for states like Bihar, Goa, Meghalaya, and Sikkim.

Similarly, the proportion of female children fetching drinking water from distant sources decreased but marginally at the national level from 2.82% in 2015–16 to 2.81% in 2019–21. This decline was majorly evident in low focus states (2.19% in 2015–16 to 1.39% in 2019–21) followed by high focus states (3.43% in 2015–16 to 3.13% in 2019–21). The same pattern of decline was also observed at individual state level except for states like Tripura, Assam, Arunachal Pradesh and West Bengal.

Apart from the person fetching drinking water, the time taken to fetch drinking water was also found to have declined significantly at the national and majority of state levels except for selected north-eastern states. At the national level, the proportion of households taking more than 30 minutes of time to fetch water reduced from 6.75% in 2015–16 to 4.87% in 2019–21. Similarly, this reduction in time consumption was observed significantly among the majority of high-focus EAG and low-focus states; however, among north-eastern states, the proportion of households taking more than 30 minutes to fetch drinking water increased from 3.75% in 2015–16 to 4.88% in 2019–21. At the individual state level, a similar pattern of reduction time consumption to fetch drinking water was evident except for Bihar, Jammu and Kashmir, and a few north-eastern states such as Meghalaya, Mizoram, Nagaland, Sikkim, and Tripura.

3.2. Factors determining coverage of household tap water connections

Our supply-induced and demand-side drinking water accessibility parameters can be determined by various socioeconomic and demographic factors. Table 2 shows the marginal effect from the probit models representing the probability of having household tap connection, water collection by adult women and water collection by female children, and time taken to fetch water among various socioeconomic and demographic groups.

In both pre-JJM and early-JJM periods, the probability of availing a household tap connection was found to be significantly high among wealthy households. The probability increased with the increase of wealth status. Compared to the poorest quintile, households belonging to the richest quintile had a 25.9% (P = 0.000) higher probability of having a household tap connection in the pre-JJM period, which marginally increased to 26% (P = 0.000) in the early-JJM period. On the other hand, the probability of water collected by women and female children and time taken to fetch water were found to be significantly high among the disadvantaged households in both pre- and early-JJM period.

Among the households who had tap water connections, the households with no education had a significantly higher probability of owning a tap water connection compared to the households with some level of educational attainment. Correspondingly, a higher probability of water being fetched by women and female children and a higher probability of longer time consumption towards fetching water was found among households with no proper education.

Among the caste categorization, households belonging to SC caste had a significantly higher probability of having household tap connections compared to the ST, OBC and other castes. At the same time, a significant proportion of households belonging to the SC caste also had a higher probability of women fetching water in their households compared to the other caste categories. However, we did not find any significant association of caste with the proportions of female children fetching water and the time taken to fetch water.

Apart from the wealth, education, and caste category, the size of the household was also found to be one of the significant determinants of drinking water accessibility. The probability of having a household tap connection was found to have significantly declined with the increase in household size. However, the probability of women and female children fetching water and a longer duration of time taken to fetch water were found to have increased with the increase in household size.

The factors determining demand and supply side indicators of access to drinking water were estimated separately for high focus EAG state, low focus states and north-eastern states and the results were similar (refer to S2–S4 Tables).

3.3. Access to tap water connections and its effect on the burden of water collection

To assess the district level effect of tap water connection on the proportion of women and female children fetching water and the time taken to fetch water, we calculated the absolute difference in the mean of these outcome indicators in each state. The difference scores of all the parameters are plotted in Fig 4. A pattern of negative association between change in the proportion of household tap connections and change in accessibility parameters can be seen in Fig 5. A positive change or an increase in the household tap connection was associated with a decline in women/female children fetching water and the duration of fetching water and vice versa. However, to examine the significance of this association, we utilized regression models.

The regression coefficients from both models are plotted in Fig 6. Our result suggests the proportion of households with tap water connections in the early-JJM period had a positive effect in terms of reducing the water collection burden on adult women i.e., districts with higher household tap connections in the early-JJM period were associated with an average of 10% decline in the proportion of household in which water was fetched by adult women. We did not find any significant association between household tap water connection and reduction of burden on female children and reduction in duration of water collection.

Apart from assessing the district-level effects of JJM across the nation, we also explored various implementation phases of JJM. We identified Har Ghar Jal (HGJ) reported villages for the year 2021 as there was no HGJ-certified village as of 2021. We attempted to estimate the effect of JJM on various indicators at the district level for states with a higher proportion of HGJ-reported villages; however, we did not find any significant effect. This negligible effect could be due to the ‘HGJ reported’ status itself. It is worth noting that ‘HGJ reported’ does not necessarily confirm a regular supply of drinking water, although it can be considered as completion of water supply infrastructure development and the claim of water supply is yet to be accepted by passing a Gram Sabha resolution [1]. The state-level association of the proportion of HGJ reported villages and the demand-side water accessibility indicators are presented in S1 Fig.

4.1 Policy recommendation

Despite the fact that overall proportion of households in which burden of water fetching on female children and duration of water fetching has reduced, our results could not attribute this reduction to the JJM coverage. There can be several reasons that need policy makers’ attention. First, mere creation of tap connections does not assure the flow of water as there can be several factors that can cause delay in the completion of a project in a particular village. Therefore, community involvement from the initial stage of the intervention is crucial to have a track of work progress and a regular and safe water flow. Second, community awareness on water utility development is crucial as people may still prefer the primitive source of water due to the variation in taste (chlorination) of tap water. Lastly, although the long persistence pro-rich bias in the coverage of tap water connection reduced in post-JJM period, it still exists. To tackle this issue, the government may collect community contributions based on the ability to pay and should also consider community contribution in kind. Even though the program targets a 100% inclusivity in its coverage, maintaining socioeconomic equity in its implementation stage will have a positive reflection in ensuring long-run sustainability.

4.2 Limitation and future scope of research

There are several limitations to our study. Our findings only indicate the early effects of JJM (till 2021) in terms of access to drinking water. This represents the initial phase of implementation in which only 22.15% of the villages were identified as ‘Har Ghar Jal reported’ and zero villages were identified as ‘Har Ghar Jal certified’ (S1 Fig). Moreover, the implementation of JJM schemes requires 12 to 18 months of time. Hence, there is a possibility that the NFHS (5^th round) data collected over the period 2019 to 2021 might have captured only 1^st and 2^nd phases of implementation in selected villages. Further, the true effect might not have been captured by the data due to the COVID-19 pandemic which disrupted most of the developmental activities.

As future scope of research, researchers may explore various aspects such as quantifying the time saved due to tap water connections and its utilization in income generation activities, the aspect of water quality (including quality testing and treatment), regularity and reliability of water supply system.