Evaluating the impact of alternate wetting and drying (AWD) technique in rice production on farm incomes and water savings in Bangladesh

Publication Details

3ie Funded Evaluation, DPW1.1081. A link to the completed study will appear here when available.

Manzoor Dar, Lutful Hassan, Ujjayant Chakravorty, Kyle Emerick
Institutional affiliations
None specified
Grant-holding institution
None specified
South Asia
Environment and Disaster Management
None specified
Gender analysis
None specified
Gender analysis
Equity Focus
None specified
Evaluation design
Randomised Control Trials (RCT), Others
Ongoing 3ie Funded Studies
3ie Funding Window
Development Priorities Window 1


The study evaluates the effect of alternate wetting and drying (AWD) technique relative to conventional flood irrigation in rice on farm incomes and water savings.


Historically, rice is grown in fields that are flooded during most of the monsoon season. However in recent years, mainly due to the availability of tube well irrigation, rice is increasingly being planted in the dry season using shallow or deep tube well water. This shift towards the use of groundwater irrigation that has occurred in many countries of South and East Asia, including Bangladesh, India, Pakistan and China. While this trend has led to large increases in food production and self-sufficiency in these nations, it has also led to an increase in groundwater extraction and depletion of water levels in many rice-producing regions.

To address this problem, researchers at the International Rice Research Institute have developed the AWD technique that can be used to save water for the purposes of helping the environment and saving for the future generations. This study will evaluate the effect of AWD relative to conventional floor irrigation in rice. Positive results will help support scale up and wider adoption of the AWD technique in the agricultural sector.

Research questions

  1. What is the average impact of AWD on farm profitability and water savings?
  2. Does AWD have external benefits by saving water or reducing emissions from either methane (CH4) or diesel fuel and electricity?
  3. How does the farmer’s revealed private benefits compare to both the costs of AWD and the estimated external benefits? 
  4. Do custom subsidies based on discount rates increase adoption relative to uniform subsidies?


Intervention design

The AWD technique works on the principle that the rice plant can tolerate up to 30 percent less water supply during the growing season relative to conventional methods of irrigation. This technique requires embedding a perforated plastic tube to monitor the water level in the rice field, which is irrigated each time the water level falls more than 15 cms below the soil surface. AWD technique was developed in 2004 and tested on farmers’ plots in Bangladesh since 2007.

Theory of change

The adoption of AWD is anticipated to translate into final impacts on crop yield (no change expected), water use (decrease), methane production from rice (decrease), weeding costs (possibly increase) and agricultural profit (increase).

The theory of change lies on few important assumptions: 1) the existence of a marginal price for water is a key incentive for adoption of AWD; 2) there is no ‘warm glow’ effect where farmers are willing to use AWD and save water just for the purposes of helping the environment and saving water for future generations; 3) farmers know how to use the AWD pipe and 4) they can only use AWD if they can obtain water when they need it.

Evaluation design

The evaluation uses a combination of a cluster randomised control trial with partial equilibrium welfare analysis to rigorously measure the impact of AWD. The experiment is split into three phases.

The first phase will estimate the average treatment effect of AWD where 400 villages will be randomly allocated to two experimental groups. Ten farmers will be selected from each village. The farmers in 200 treatment villages will be provided with the AWD pipe, instructions on its benefits and training in using it to measure soil moisture. The farmers in the remaining 200 villages will act as a pure control. The ten surveyed farmers in these villages will be given AWD equipment at the end of the first year to observe the plot on which these farmers choose to use AWD in the second year.

The second phase will estimate the private demand for AWD. The AWD equipment will be offered for sale to a new random group of ten farmers in each of the 200 treatment villages and 60 randomly selected control villages from the first phase. These villages will be split into four groups and the price of AWD equipment randomised at the village level. Prices can vary to equal 1) the cost of acquisition; 2) the production cost of each AWD pipe and the remaining two prices within this range.

The third phase will explore whether subsidies for AWD can be precisely targeted. A new set of 180 villages in the same area will be randomised into three different groups of 60 villages each. Farmers in the first group will pay a uniform subsidised price for AWD based on estimates from the second phase. In the second group, they pay a price based on their discount rate. For the third group, the subsidised price is based on observable correlates of demand elasticity from the second phase.

Scroll to Top