Farmers’ Willingness to Engage in Best Management Practices: an Application of Multiple Imputation

The purpose of this article is to explore how much farmers may engage their lands in Best Management Practices (BMPs) through a water quality trading (WQT) program in Kentucky. We conducted a survey of farmers in the Kentucky River watershed from 2011 to 2012. The questions in our survey are whether and how much farmers may adopt the BMPs (in addition to what they have already used) if they are offered compensation through WQT. About 20% of respondents with respect to five different types of BMPs did not indicate how much they will adopt. We compare three strategies to handle the missing data: deleting the observations with missing value, simple imputation by imputing the mean of the observed value, and Multiple Imputation (MI). Follow these missing data treatments, we estimate the factors affecting how much farmers may engage in BMPs using Tobit model. The results show that increasing the compensation for using BMPs are more likely to encourage farmers to adopt riparian buffers, animal fences and nutrient management. Besides, land area, rent area, farming revenue, percentage of reinvestment to farms from household income, water quality near the farm, participation in government programs, and current experience of BMPs will affect BMPs adoption. The results by using the MI are more promising and reasonable than using the deletion method and mean imputation method.