A Predictive Model for Farmland Purchase/Rent Using Random Forests

This study contributes to guidance for understanding farmland purchase and rent decisions in Korea via an analysis using a machine learning tool, Random Forests: A Supervised Machine Learning Algorithm. Farm Household Economy Survey is employed to predict the relationship between farmland acquisition and farm household economic characteristics. Our main findings are two folds. First, a farmland purchase decision is positively related to transfer incomes, the value of inventory & fixed assets, and the value of farmland that farmers owned. Second, a farmland rent decision is also positively associated with a rent paid in a prior year, revenue from field crops, inventory and agricultural assets, and transfer incomes.