Mission Kakatiya is a minor irrigation project under the Department of Irrigation and Command Area Development of the Government of Telangana (GoT), India, involving civil works to repair and restore water tanks in the state for local agricultural water use. The government plans to roll out this project to all tanks in the state—approximately 46,500 in total—over five phases spanning 5-7 years.

Using this context, the researchers present a first-of-its-kind evaluation that combines experimental and observational methods to evaluate the impact of increased access to irrigation on local agriculture and water markets.

Why this study

Tank rehabilitation is increasingly becoming central to climate adaptation and water management policies in India as evident from the growing number of such rehabilitation projects. Tank irrigation projects are more cost-effective than medium- and large-scale irrigation projects. The restoration work under Mission Kakatiya is for all the tanks across Telangana. Hence, an evaluation of the impact would be useful for deciding on a policy for ongoing maintenance, given that the GoT has already committed huge sums of money for rehabilitation, but none yet on maintenance.

Additionally, a systematic evaluation can provide valuable insights to other southern states that account for 60 percent of the 3.2 million hectares of tank irrigated area in India.


The research entails a two-part impact evaluation of the program. It investigates how a tank restoration effort by the Central government impacts crop choice (paddy, cotton, and maize), farm productivity, agricultural output for important crops (paddy, cotton, and maize), water table levels, and agricultural income (revenue net of costs) in the area surrounding a tank.

The study examines corollary questions such as identifying the characteristics of tanks that have the largest impact on primary outcomes and determining how water from tanks is distributed among farmers in the area served by the tank.

The first component of the evaluation involved a non-experimental difference-in-differences methodology, comparing tanks undergoing rehabilitation under phase 1 of the program with phase 2 and those that were yet to be rehabilitated as of 2017.

Researchers identified a sample of 752 across these groups, stratified by Assembly Constituency (the lowest unit where decisions are made with regard to which tanks are to be selected under a specific phase), using a propensity score matching method. Phase 1 tanks, most of which were restored as of summer 2016, served as treatment tanks, and phase 2 tanks (most of which were yet to be restored as of early 2017) and other tanks yet to be rehabilitated, served as control tanks. The study sampled 262 tanks from phase 1, 318 from phase 2 and 119 non-rehabilitated ones in 32 mandals (administrative divisions) across the former districts of Mahbubnagar and Warangal. We sampled 5 farmers per tank with 2 near the head-end and 3 near the tail-end of the tank command area (ayacut). When a village had more than one tank, we randomly selected one tank to choose our sample of farmers. The data collection for the non-experimental component of the study is complete, and the data analysis is currently underway.

The second component (ongoing) assesses tank rehabilitation under phase 4 through a randomized control trial (RCT). The sample for this component consists of 92 tanks across 15 newly formed districts in the state of Telangana. Of 92 tanks, 46 were randomly assigned to be rehabilitated in phase 4 (during 2018), with silting completed before the monsoon, whereas rehabilitation of the remaining 46 (assigned as control) has been delayed for later phases.

To compute the effect on farm-level outcomes, the study sampled about 30 farmers who cultivate within the area served by a tank, covering locations across head, middle and tail end. The study sampled entire population of farmers when their numbers were less than 30. Otherwise, a random sample of 30 farmers was drawn among the total population within the tank area, stratifying by location—head, middle, and tail end.

Preliminary Findings

  1. An increase in volume of water available among rehabilitated tanks compared to non-rehabilitated tanks, though the difference is not statistically significant. However, since all groups recorded an increase in volume, the study may have to account for good monsoons since rehabilitation commenced.
  2. Among a sub-sample of tanks where there was no irrigation in 2015 (pre-period) due to non-availability of water, there is a significant increase in irrigated areas among phase 1 and 2 tanks compared to non-rehabilitated tanks, but no difference between phase 1 and 2.
  3. Little difference between rehabilitated and non-rehabilitated tanks with respect to dry season water levels and volume of water in storage.
  4. A greater fraction of surveyed farmers used tank water for irrigation in phase 1 and 2 areas compared to non-rehabilitated areas post the program.
  5. Greater number of days of tank irrigation among farmers in rehabilitated areas.


While the study finds modest effects of tank rehabilitation on immediate outcomes in the observational study, these could be driven through unobserved differences in trends between the yet-to-be rehabilitated and non-rehabilitated tanks. The second experimental component will enable us to confirm whether these observed patterns can be causally attributed to the rehabilitation program.