Employment generation potentials in the implementation of a rural drinking water supply program: evidence from India's Jal Jeevan Mission

With the unemployment rate hovering at 8% and youth unemployment at 10%, Indian policymakers are incessantly exploring avenues to generate employment (CMIE, 2024; ILO, 2024). The task has been challenging as it requires creating employment for the current unemployed persons as well as 7–8 million newer young people joining the labour force every year (ILO, 2024). The role of large-scale infrastructure developmental programs/schemes can have catalyzing multifaceted effects on direct and indirect employment generation (Parikh et al., 2018; Bennett, 2019; Ngulube, 2022; Nourelfath et al., 2022). With poor rural infrastructure, the slightest improvement in basic facilities like roads, electricity, telecommunication, housing, health, water, and sanitation can have a significant impact on economic development and human well-being in India (Ghosh, 2017). The year 2019 was a watershed year in the history of drinking water supply in rural India with the launch of a large-scale rural development program, the Jal Jeevan Mission (JJM). This program aims at providing tap water connection to individual households (HHs) with the provision of 55 L of potable water per capita on a daily basis. Since its launch in 2019, the program has managed to cover 3/4th of rural HHs in the country. Some of the primary objectives of the program are to provide a regular supply of adequate quantity of potable water to prevent deaths and illness due to waterborne diseases, eliminate drudgery in accessing drinking water, and improve the health and productivity of people in rural areas. One of the distinct features of the mission is to focus on service delivery rather than just creating infrastructure. Moreover, with the scope of decentralized governance and a greater community engagement, the mission holds significant potential for spillover effect in generating employment in various phases of its implementation.

The past literature throws light on the impact of infrastructure development (access to basic amenities such as drinking water, toilet facilities, electricity, and housing) on economic indicators such as Gross Domestic Product (GDP) per capita (Munnell, 1990; Sahoo et al., 2010; Dwivedi, 2017). Development activities in the construction sector are mostly reported to have a strong positive effect in addressing the problem of cyclical unemployment (Ball, 1965). Furthermore, the program-based approach of public work was stressed above the project-based approach in many developing countries due to its longer-term nature and ability to address the unemployment problem (Thwala, 2008). Ascribing to the benefits of the program-based intervention approach, India has implemented several developmental programs that had either intentionally or unintentionally accelerated employment generation in the country (Nayak & Hazarika, 2019). Studies assessing the employment generation potential of government policies were highly skewed towards the schemes related to housing, roads, health, and education (Ball, 1965; Strassmann, 1976; Tipple, 1994). However, it is crucial to note that the construction and maintenance of water supply schemes, when locally managed, possess the potential of creating jobs at the lowest economic level where unemployment tends to be high with lower skill levels (Wall, 2023). For instance, under the JJM, the creation of infrastructure such as functional household tap connections (FHTCs), water storage tanks, and water treatment plants provides direct employment to labourers for construction activities and laying pipelines, whereas skilled workers such as engineers, valve men, pump operators, and managerial staffs are directly employed for proper execution of the schemes. Similarly, the maintenance of the scheme also requires several types of skilled manpower, such as watermen, pump/valve operators, supervisors, and watchmen, to regularly inspect and ensure uninterrupted service delivery. So far, the literature provides sparse evidence on how quality drinking water facilities can lead to better health, higher labour supply, and productivity (Asit et al., 2005; Kremer et al., 2011; Devoto et al., 2012), but no proper evidence of employment generation as an effect of implementation of rural water supply program. Against this backdrop, our study aims to assess the employment generation potential of the JJM program in the construction as well as operation and maintenance phases of its implementation.

We estimated employment potentials under JJM using two different methods. First, we utilized the input–output (IO) model to estimate the overall employment potential (direct + indirect) during the construction phase. Subsequently, using the ratio method, we estimated the direct employment potential in different phases of the implementation of JJM schemes. Furthermore, we also estimated the partial indirect employment generated during the construction phase, which involves the production, storage, and transportation of materials using the ratio method. We did not attempt to compute the indirect employment created during the O&M phase and other induced employment due to data constraints. To make this study nationally representative, we used data from selected states across the country and estimated the employment potential of other states using cluster analysis.

We utilized the IO table published by the Asian Development Bank (Asian Development Bank, 2023) to derive the Leontief inverse matrix. Since rural drinking water supply (RDWS) is not identified as one of the industries in the current IO table, we created a synthetic industry to study the impact of investments in JJM. The output created under piped drinking water is already a part of identified industries in IO tables and we constructed the industry ‘RDWS’ as an aggregate of identified industries. For the creation of the synthetic industries, we used data from 11 tenders across four states (Karnataka, Odisha, Himachal Pradesh, and Kerala). The contracts belong to different phases of the scheme and range from tender values of INR 0.4–5 million. These contracts were obtained from the e-procurement portal of each of these individual states. Each line item in the tenders was studied and classified under an identified industry. The combined sum of costs of all the contracts was used to calculate the percentage contribution of the existing identified industries to the synthetic industry. The breakdown of the share of the synthetic industries is summarized in Supplementary Table 1.

The demand vector is generated by multiplying the total likely investment in JJM with the share of the component industries (Supplementary Table 2). The likely investment is computed by multiplying the sample average expenditure per HH and the total number of HHs to be covered.

The employment output ratios have been derived for the seven major industries laid out in the national income accounting data. Since the industry-wise allocation of GDP was not available, we used the percentage allocation of Gross Value Added (GVA) calculated for the year 2022–2023 and applied it to the GDP data. We believe that this is the closest estimate possible since the GDP of a country is net taxes added to GVA.

The total worker population for the country stood at 508,265,520 at the end of 2022. This was derived by multiplying the worker population ratio (WPR) for the year 2021–2022 with the total population (age >15 years) estimated by the World Bank for 2022. The WPR is calculated based on the Periodic Labour Force Survey carried out by the National Sample Survey Organisation every year. The worker population is further allocated to the eight major industries as per the allocation of GDP and the GDP per worker is computed (Supplementary Table 3). We then mapped these eight employment output ratios to 35 industries in the IO table based on their correspondence, as indicated in Supplementary Table 4.

For estimating the direct employment generation under JJM schemes, we considered the total employment generated and the total number of HHs covered at the scheme level from the data collected from various states. For cluster analysis, various state-level parameters related to demography and water availability were collected from the Economic and Political Weekly Research Foundation (EPWRF) – India Time Series (EPWRF, 2023). Finally, the employment-to-HH ratio normalized to 100 HHs was calculated.

For the direct employment estimation, we collected scheme-level data from various states. For this, we selected one highest scoring (best performing) district from each region of all major states based on the Jal Jeevan Survekshan report of December 2022 (Government of India, 2022). We listed all the completed schemes in the selected district and requested the JJM directors of the respective states to provide scheme-level information. A data format with a list of sample districts and a list of completed schemes (n = 854) was shared with the respective states. Details of sample districts and number of schemes are given in Supplementary Table 5. The data format included questions related to the scheme characteristics and employment type, such as nature of the scheme (new/retrofitting), category of the scheme (SVS/MVS), phase of implementation (construction/O&M), number of villages covered, number of population and HHs covered, estimated cost of the scheme, total water supply capacity, service-level capacity, and a set of questions related to employment types. Furthermore, we reached out to eight contractors from Tamil Nadu and four contractors from Karnataka and interviewed them using the same sets of questions. For the sake of robustness, we considered employment units as ‘person-years of employment’, which takes into account both the number of persons employed and the amount of time each person dedicated. Since the program is currently under implementation, the total number of completed schemes was dynamic in nature. Due to this, we received surplus data from 1,067 schemes.

For the clustering of schemes, we considered state-level parameters such as population density, river length per 1,000 population, water body area per 1,000 population, groundwater availability per 1,000 population, and WPR of casual labour per 1,000 population to generate the clusters. We used ‘K-means’ clustering and employed Calinski–Harabsaz (CH) pseudo-F statistics to plot the elbow chart. The CH pseudo-F assesses the sum of squared distance within the cluster and compares it to the un-clustered data, taking into account the number of clusters (Halpin, 2016).

The values of CH pseudo-F are then plotted against the number of clusters (k) to identify the kink/elbow point on the curve, which denotes the optimal number of clusters (k), which is 4 in our case (Supplementary Figures 1 and 2). As the fourth cluster had only one state, i.e., Arunachal Pradesh, we clubbed it with the third cluster, which had other major northeastern states. The three clusters of the states are presented in Tables 3 and 5. Furthermore, due to the unavailability of data, we did not include Union Territories (UTs) in our cluster analysis; however, we created a separate group for UTs for which employment is estimated using national average ratios.

We utilized the Leontief IO model to estimate the overall employment potential, which includes direct and indirect employment generation during the construction phase. The IO model in our study was adopted from Garrett-Peltier's study, in which the impact of additional investments in the renewable energy industry was estimated on employment (Garrett-Peltier, 2017).

To derive the impact on employment, we arrive at an employment requirement matrix (E_r) from the Leontief inverse matrix and the employment requirements coefficient matrix (E), where E is a diagonal matrix indicating the employment output ratios (number of people employed/ total output) for each industry. The matrix E_r helps us estimate the number of jobs generated, both directly and indirectly, at any level of planned output.

We estimate the employment generated by JJM investments using Equation (3). We generated the employment requirements coefficient matrix (E) by computing the ratio of employed persons to output for each component industry of drinking water supply, which is then used to arrive at the employment requirement matrix (E_r).

The employment to HH ratio representing employment generated per HH was computed using a sample of schemes from selected states. Subsequently, the estimated ratios were utilized to extrapolate the results at the state and national levels. As we know the total number of HHs to be covered under JJM, we consider ‘HH’ as the unit of estimation. This method is used for direct employment estimation in both construction and O&M phases.

To compute the potential employment in states other than the sample states, we computed the average DLE in each cluster and used it to compute employment in states other than the sample states within the respective clusters.

The same approach is followed in the O&M phase to estimate the employment generation potential utilizing scheme-level sample data.

We computed output generated per employment of each industry (average value of output generated by an employed person) and then arrived at the employment generated due to the additional demand of the input materials used for JJM schemes by multiplying it by BmM_k. The total indirect labour employment estimated through this method is partial as it captures only the first stage of indirect employment.

Tables 1 and 2 show the summary statistics of the scheme and their characteristics. A total of 1,067 scheme-level data were collected, among which 81 schemes belonging to Assam and Goa and 69 schemes from other states were dropped after screening for outliers and population coverage of at least 20 HHs and 100 people. As a final consideration, we included 917 schemes, out of which 58.01% are SVSs, 41.98% are MVSs, and 72.08% are from the construction phase and 27.91% are from the O&M phase. Furthermore, among the schemes from the construction phases, 76.18% were retrofitted and 23.81% were new schemes.

The scheme coverage statistics show that on average, the total manpower employed under an MVS is 26 and SVS is around 4; total villages covered for MVS and SVS are 9 and 1 with an average population of 15,874 and 1,093 people and 2,677 and 278 HHs, respectively. The average water supply capacity per MVS and SVS is 1.85 and 0.49 MLD (million litres per day) with an average per capita cost of INR 8,445 and 7,549, respectively.

The IO model estimates the direct and indirect employment potential across both MVS and SVS schemes. The total estimated investment under JJM was obtained by multiplying the sample average cost of HH connection and the total number of rural HHs to be covered. The estimated overall employment generated at the construction phase obtained from the IO model is 28,248,478 person-years for the total estimated investment of INR 7.80 trillion under JJM (Supplementary Table 4).

The construction industry has the highest employment potential at 13,942,573 person-years followed by the machinery and basic metals industry. This is due to the nature of the water treatment plants and distribution networks, which require heavy investments in constructing large tanks, large amounts of iron and steel pipes, etc. The construction industry has 49% contribution to employment generated but only a 35% contribution to the increased output, as the employment intensity of the industry is higher than the overall average.

Whenever investments in large infrastructure projects are made, there is a multiplier effect on the economy. In the construction phase, the employment used while constructing the infrastructure is considered direct employment under JJM, the employment generated to produce the materials used in the construction is the first stage of indirect employment, and the employment generated in producing raw materials for the first stage is the second stage indirect employment, and so on. Using the IO model, we get the aggregate employment potential across multiple stages. To break this down into the direct employment potential and the first stage of indirect employment, we used the ratio method.

The direct employment generated under the construction of schemes at the state and national level is provided in Table 3. Overall, JJM has the potential to generate 5,993,154 person-years of direct employment in the construction phase of the water supply schemes. The highest ratio of employment generation in the construction phase was in Maharashtra (6.31 per 100 HH) followed by Tamil Nadu (4.40 per 100 HH), whereas the ratios of employment generation for Andhra Pradesh (0.4 per 100 HH) and Gujarat (0.6 per 100 HH) are the lowest. This variation is mainly due to the differences in the type and nature of the schemes.

To calculate the indirect employment generated by JJM, we have attempted to estimate the breakup between spending on materials and labour. The average proportion of material cost across all the sources was 72%, which is indicative of the total tender budget used on physical materials, the breakdown of which is summarized in Appendix 5. Considering the output from these industries is generated at the average productivity of employed persons in India (Supplementary Table 6), the indirect employment generated in the first stage stands at 7,734,620 person-years during the construction phase of the mission. The employment generated in specific industries is summarized in Table 4. The remaining 14,520,704 person-years of employment, out of the total indirect employment of 22,255,324 person-years (overall employment in construction phase − direct employment in construction phase), is generated in the production of inputs used in manufacturing of materials required in the first stage.

As shown in Table 5, by 2019, there were 32,362,838 rural HHs with FHTC; furthermore, a total of 162,217,522 HHs were planned to be covered under JJM. At the national level, JJM is potentially generating 1,325,919 person-years of employment in the O&M phases; out of which 1,118,749 person-years of employment can be ascribed to the JJM period (after 2019).

To estimate different types of employment, we created two categories: (i) skilled labour employment and (ii) unskilled labour employment, estimated as a part of direct employment using the ratio method. The unskilled labour employment includes helpers/watchmen and labourers, while the skilled labour employment takes into account the remaining positions. In some cases, data did not reveal the type of employment. We included them under unskilled employment.

Our result suggests that JJM has the potential to generate 2,620,783 person-years of skilled labour employment (43.72%) and the remaining 3,372,371 person-years of unskilled labour employment at the national level in its construction phase (Supplementary Table 7). Meanwhile, in the O&M phase, JJM has the capacity to generate a total number of 729,156 person-years of skilled (65%) and 389,593 person-years of unskilled labour employment annually (Supplementary Table 8).

Although it is difficult to assess the long-term impact of the JJM at this point of time, the program has a huge scope for short-term evaluation as there are no proper studies assessing the immediate and intermediate effects of JJM. The present literature indeed provides some evidence related to the conceptual framework of JJM (Sarkar & Bharat, 2021; Bongirwar & Dahasahasra, 2022; Singh et al., 2023) and its crude coverage (Wescoat et al. 2022). However, there are no proper studies assessing the ground-level impact of the program except one which assessed the social, health, and economic impacts with the help of a primary survey in rural regions of Karnataka and found significant positive effects of JJM (Premalatha, 2023).

Our study has several caveats. First, the program is still in its implementation phase, and the total number of schemes at any point of time was dynamic; hence, it was difficult for us to draw a sample of schemes with minimum frame error. Moreover, this is one of the impeding factors that constrained us from estimating employment separately for different scheme types (MVS/SVS) and nature (new/retrofitting). Second, our study does not capture the induced employment effect in the indirect employment creation due to the unavailability of data. Furthermore, due to data constraints, indirect employment estimation was considered only for construction phases. Third, although we estimated skilled and unskilled employment, the availability of data did not permit us to assess the quality of work and identify the beneficiaries at the ground level. Nevertheless, to the best of our knowledge, our study is a novel attempt to assess the potential positive gain due to the implementation of a publicly funded rural water supply program.

To encapsulate, the study findings reveal a significant impact of the JJM in terms of creating employment in the economy. Being rolled out as a national-level program, the mission aims not only to make access to safe drinking water easy but also to emerge as an opportunity hub creating short-term as well as long-term employment. We estimated the total employment likely to be generated due to the implementation of JJM. Our findings call for the attention of policymakers towards the proper execution of schemes in the post-implementation phase, elucidating the provision of manpower requirements in the operational guideline. Moreover, components such as decentralized governance and community involvement have to be given priority, which will play a vital role in making employment creation in the maintenance phase long withstanding.

We acknowledge the valuable suggestions received from the International Labour Organization team led by Tomas Stenstrom. We thank the state governments and the respective mission directors of Andhra Pradesh, Gujarat, Karnataka, Kerala, Punjab, Uttar Pradesh, Maharashtra, Tamil Nadu, Uttarakhand, and Jharkhand for furnishing the required data that helped us conduct this study.

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

The authors declare there is no conflict.