What Works for Water Conservation? Evidence from a Field Experiment in India

Many studies have demonstrated the effectiveness of promoting yield-enhancing technologies among agricultural producers in the developing country context. However, resource-saving technologies, which provide more long-term as opposed to short-term benefits, have been less studied in this context. In a randomized field experiment we test whether informing medium scale farmers in Haryana, India, reduces water use from common aquifers. We find that training farmers in an agricultural water reduction practice (alternate wetting and drying), conditional on whether we randomly metered a farmer's pump, reduced farmers' self-reported pumping hours during peak pumping weeks, and strengthened farmers' beliefs in potential farmer group cooperation. In addition, we also test whether incorporating social comparison messages about a farmer group's average pumping hours impacts use given the incentive to free ride. We find no added effect of these messages when coupled with training. Our main results are robust to various specifications including the use of double/debiased machine learning, which can correct for potential multicollinearity of confounders while still obtaining unbiased estimates of our treatment effects