Abstract

The objective of this work was to quantify technical efficiency in 1086 dairy farms from Costa Rica (year 2007) using Stochastic Frontier Analysis. The base model analyzed the endogenous variable logarithm of Milk Solids (kg/ha/week) as a function of 14 different predictive variables related to physical and management aspects. Overall mean for milk solids was 32.5±23.5 kg, with the highest mean for typology of Specialized Intensive Dairy Herds from the Highlands (83.9 kg), followed by Intensive Dairy Herds from the Lowlands (51.0 kg), Semi-Intensive Dairy Herds from the Highlands (33.8 kg), Extensive Dairy Herds from the Lowlands (23.0 kg) and Dual Purpose Herds from the Lowlands (8.5 kg). Variables associated (P<0,001) to milk solids were stocking rate, fertilizer application, concentrate feeding, proportion of cattle from specialized dairy breed, altitude and grazing area. An increase of 1% in these variables was associated with a change of 0,85%, 0,07%, 0,07%, 0,19%, 0,15% and -0,10% in milk solids, respectively. Overall technical efficiency was 0,75±0,09. When farm typology was added as a predictor into the model, efficiency increased to 0,79±0,07, whereas adjusting the model separately for each farm typology resulted in an overall efficiency of 0,77±0,11. The latter was preferred because better reflects heterogeneous elasticity of predictive variables between different typologies.

Keywords: stochastic models, Cobb-Douglas, production frontier.