Revitalizing Traditional Water Tanks for Climate Adaptation in India

There are two kinds of interventions involved:

1) Tank renovation: villages could have two kinds of structures - tanks, which are like man-made lakes/reservoirs, and ‘dohas’, which are saucer-shaped structures dug in the stream bed along its length separated into multiple, smaller water ponds by mud dams to prevent overflow. The water tanks/dohas were originally built at the foot of a hill or a natural slope to capture rainwater during the monsoon (June-September) and they formed the main source of stored water in these villages. Currently, after accumulating silt over 1000 years, the tanks/dohas have reduced volumes of water and tend to run completely dry by February/March, restricting farmers’ ability to crop and irrigate all of their cultivable lands in the dry season, and limiting the water available for domestic activities and livestock rearing. SRIJAN is engaged in renovating these tanks by digging out 8-10 ft of the silt across the whole structure which will increase its carrying capacity such that the water could last longer, ideally until the next monsoon hits.

2) Training: The training intervention involves repeated sessions with the NGO’s experts before and during the dry season (November to May) that cover different aspects of agricultural practices that can improve yields, make the crop robust to shocks, and enhance water efficiency on the farm.

There are two kinds of interventions involved:

1) Tank renovation: villages could have two kinds of structures - tanks, which are like man-made lakes/reservoirs, and ‘dohas’, which are saucer-shaped structures dug in the stream bed along its length separated into multiple, smaller water ponds by mud dams to prevent overflow. The water tanks/dohas were originally built at the foot of a hill or a natural slope to capture rainwater during the monsoon (June-September) and they formed the main source of stored water in these villages. Currently, after accumulating silt over 1000 years, the tanks/dohas have reduced volumes of water and tend to run completely dry by February/March, restricting farmers’ ability to crop and irrigate all of their cultivable lands in the dry season, and limiting the water available for domestic activities and livestock rearing. SRIJAN is engaged in renovating these tanks by digging out 8-10 ft of the silt across the whole structure which will increase its carrying capacity such that the water could last longer, ideally until the next monsoon hits.

2) Training: The training intervention involves repeated sessions with the NGO’s experts before and during the dry season (November to May) that cover different aspects of agricultural practices that can improve yields, make the crop robust to shocks, and enhance water efficiency on the farm. This will cover pre-sowing information on crop variety selection, land preparation, sowing techniques, and seed treatment; irrigation management and water budgeting techniques; weed and pest management techniques; organic fertilizer processes; crop growth monitoring and assessment, harvesting guidelines, and post-harvest management. The training will focus on information that is relevant for that year considering the expected temperatures and the level of rainfall in the monsoons. Modules like crop choice, sowing practices, irrigation management, water-budgeting will be especially important to discipline water use.

2023-06-01

2025-07-01

Our outcomes of interest will also span tank-level outcomes:

Water levels across different months captured through farmer reports in survey data as well as satellite imagery; water-source level outcomes: water-levels in wells owned by sample farmers across different in-person survey time points

Household-level outcomes: crop choice, irrigation practices, yield, profitability, consumption

Input-market outcomes: the price of water, and the going agricultural wage for different tasks.

Household-level outcomes: family versus hired labour inputs for different farm tasks broken down by gender, temporary migration to urban areas during the lean season, women’s time use - especially for domestic water collection

This project evaluates the individual and complementary effects of public and private climate adaptation tools for farmers in the context of publicly managed water reservoirs and training targeted to farmer’s individual investments. These villages could have two kinds of structures - tanks, which are like man-made lakes/reservoirs, and ‘dohas’, which are saucer-shaped structures dug in the stream bed along its length separated into multiple, smaller water ponds by mud dams to prevent overflow. The water tanks/dohas were originally built at the foot of a hill or a natural slope to capture rainwater during the monsoon (June-September) and they formed the main source of stored water in these villages. Currently, after accumulating silt over 1000 years, the tanks/dohas have reduced volumes of water and tend to run completely dry by February/March, restricting farmers’ ability to crop and irrigate all of their cultivable lands in the dry season, and limiting the water available for domestic activities and livestock rearing.

Villages were randomized into treatment groups by stratifying on state and whether they have a tank or a doha, since the effects of a doha may be more locally concentrated relative to a tank. Our design is a 2X2 RCT with the following treatment groups:
Treatment 1: 39 villages. Tanks/Dohas renovation only.
Treatment 2: 39 villages. Training only.
Treatment 3: 38 villages. Tank/Doha Renovation + Training.
Control group: 38 villages. Neither.

This research is a Randomized Control trial where the intervention implementation is wholly undertaken by the NGO partner - SRIJAN, and research implementation is conducted by JPAL South Asia. Our sample covers 154 villages in the Bundelkhand region that spans two states in central India - Madhya Pradesh and Uttar Pradesh. These 154 villages were identified by Srijan as eligible to receive their interventions after conducting a needs assessment and locating a historical water reservoir.

These villages could have two kinds of structures - tanks, which are like man-made lakes/reservoirs, and ‘dohas’, which are saucer-shaped structures dug in the stream bed along its length separated into multiple, smaller water ponds by mud dams to prevent overflow. The water tanks/dohas were originally built at the foot of a hill or a natural slope to capture rainwater during the monsoon (June-September) and they formed the main source of stored water in these villages. Currently, after accumulating silt over 1000 years, the tanks/dohas have reduced volumes of water and tend to run completely dry by February/March, restricting farmers’ ability to crop and irrigate all of their cultivable lands in the dry season, and limiting the water available for domestic activities and livestock rearing. SRIJAN is engaged in renovating these tanks by digging out 8-10 ft of the silt across the whole structure which will increase its carrying capacity such that the water could last longer, ideally until the next monsoon hits.

The villages were randomized into treatment groups by stratifying on state and whether they have a tank or a doha, since the effects of a doha may be more locally concentrated relative to a tank. Our design is a 2X2 RCT with the following treatment groups:
Treatment 1: 39 villages. Tanks/Dohas renovation only. 24 completed in Year 1, 15 to be completed in Year 2
Treatment 2: 39 villages. Training only.
Treatment 3: 38 villages. Tank/Doha Renovation + Training. 22 completed in Year 1, 16 to be completed in Year 2
Control group: 38 villages. Neither.

Other than the removal of silt, the tank/doha renovation intervention involves any pending repairs to dam structures, the constitution of a tank management committee that consists of all households within the structure’s catchment area with elected leaders, and the option for farmers to take the excavated silt back to their farmlands to improve soil quality if they can pay for its transport. The distribution of the silt is overseen by the tank management committee to ensure that everyone who can pay for the silt transport gets some for their farm, which may involve rationing. Tank management committees are an integral part of the intervention whose role is to overcome the collective action problem.

Anecdotally, SRIJAN has mentioned that once a tank is renovated, farmers' initial response is to extract large amounts of water and make water-intensive decisions which leads to greater capture of the water by a few farmers, and faster depletion of the tank water levels. In their experience, it takes multiple years for farmers to adopt a conservative approach to using the water and fostering cooperation imposed by the tank management committee before benefits are reaped more widely and sufficient reserves of water are sustained through the summer. We will test if the training intervention will help establish this conservative equilibrium sooner through our experiment.

The training intervention involves repeated sessions with the NGO’s experts before and during the dry season (November to May) that cover different aspects of agricultural practices that can improve yields, make the crop robust to shocks, and enhance water efficiency on the farm. This will cover pre-sowing information on crop variety selection, land preparation, sowing techniques, and seed treatment; irrigation management and water budgeting techniques; weed and pest management techniques; organic fertilizer processes; crop growth monitoring and assessment, harvesting guidelines, and post-harvest management. The training will focus on information that is relevant for that year considering the expected temperatures and the level of rainfall in the monsoons. Modules like crop choice, sowing practices, irrigation management, water-budgeting will be especially important to discipline water use.

Randomization done in office by a computer

Randomization at the village level

Yes

154

1232 (8 per village)

Treatment 1: 39 villages.
Treatment 2: 39 villages.
Treatment 3: 38 villages.
Control group: 38 villages.

We use data from a recent village clustered RCT that we did with Indian rice farmers (Emerick et al., 2016). The plot-level crop yield is 2,817 kg per ha with a standard deviation of 1,354. The intra-cluster correlation (ICC) is 0.3449. This is high for the average cluster RCT, but not for agricultural outcomes which tend to be more correlated within villages. For this reason, our RCT uses 150 villages, rather than a smaller number of villages with more farmers per village. The strata fixed effects and baseline yields explain 17.9 percent of the variation in yield. We assume that we will obtain similar predictive power with our stratification and conditioning on the baseline outcome. Finally, we assume that each farmer will have 1.5 plots in the potential command area of the tank, making 12 observations per village. Using these parameters, we estimate a MDE of 433 kilograms per hectare on the pooled treatment effects. This is 0.32 standard deviations or about 15 percent of the mean yield. This seems like reasonable MDE, particularly since irrigation can be transformative. For instance, Jones et al (2021) find that irrigation increases cash profits by about 50% in Rwanda.