This research introduces a novel approach to evaluate energy utilization in agribusiness within a cost-based framework, addressing the complexities associated with uncertain input energy content. The study focuses on the agricultural sector in Latin America, with particular emphasis on banana plantations, examining the sector's growing dependence on energy-intensive methods and inputs. We differentiate between technical and allocative efficiency, employing a frontier-based methodology. The analysis encompasses data from 21 state-owned banana plantations in Latin America, revealing significant room for improvement in energy efficiency. Our findings indicate an average Technical Efficiency of 0.69, Allocative Efficiency of 0.91, and overall Energy Efficiency of 0.59, suggesting potential reductions in input and energy consumption of up to 38%. The study underscores the importance of addressing both input mismanagement and misallocation to enhance energy efficiency. While the proposed methodology offers valuable insights for public sector enterprises, particularly in agriculture, further research is necessary to extend this approach to scenarios with uncertain input energy content, potentially incorporating stochastic frontier analysis or fuzzy logic methodologies.