This thesis measures the efficiency of individual Agricultural Credit Associations (ACAs) with and without non-performing loans (NPLs) included as an undesirable output using a slacks-based measure based on the directional technology distance function (DDF-based SBM). The results show that first, the average efficiencies across ACAs by period measured by models with and without NPLs are statistically different. Second, the production efficiency of an association is correlated with its size and continuity (e.g., M&A or exit). Third, the estimated efficiency scores of the individual associations in a single period identify the performance of each type of input, good output, and bad output, which can help managers learn about the production performance of their association and improve association performance.