The estimation of production functions usually relies on the assumption that all producers use the same technology. However, it is possible that some firms use different technologies. In this paper, we use a stochastic frontier latent class model to estimate the technology of several dairy farms’ production systems which present different levels of intensification. The results of this model are compared with two other methods: the classical stochastic frontier model; and the estimation of separate production functions based on the classification of farms using a cluster algorithm. The empirical analysis uses a balanced panel of 169 dairy farms over the period 1999 to 2004.

Key words: Dairy farms, production function, stochastic frontier, unobserved heterogeneity, latent class model.
JEL classification: Q12, C23.