This article introduces a dynamic stochastic frontier analysis (SFA) framework with unobserved heterogeneity, persistent and transient inefficiency effects, based on recent advances in the SFA literature. The newly developed dynamic frontier model is applied on a sample of French crop-livestock farms. The estimates provide useful insights for the estimation of the technical efficiency scores as well as for the analysis of the associations of contextual drivers, such as public subsidies and indebtedness, with technical efficiency.