Effects of training duration and the role of gender on farm participation in water user associations in southern Tajikistan: implications for irrigation management

This paper examines whether longer training increases farm participation in community-managed water user associations, in a context where assignment to training duration was not randomized and none of these institutions existed before training began. We also examine whether participation is affected when farm managers migrate and leave farm operations to other workers, in a context where only managers have been directly trained, almost all managers are male, and females are increasingly operating farms. We collected microdata from 1855 farms in Southern Tajikistan, where farm managers in 40 subdistricts received longer training, while those in the other 40 received shorter training. These ‘treatment’ and ‘control’ subdistricts were selected by constructing propensity scores and matching without replacement to address observable selection effects that may affect assignment to training duration. Farms were then selected from a census using a stratified random sampling process. A difference-in-difference technique with right-hand-side covariates is employed, where both sets of data were collected after training was completed. This choice of econometric methods controls against farm-level selection effects, but introduces a potential bias due to measurement error. Longer training has a causal effect on increasing participation in WUAs. Results also demonstrate that when male workers not directly trained operate farms, participation is not affected; however, participation is negatively affected when female workers operate farms. These results provide evidence for designing irrigation management programs to target female workers directly, in order to strengthen institutions whose success depends on active farm participation.