The transposition of this methodology to the field of education is relatively straightforward in theoretical terms but faces important empirical difficulties. They concern, in the first place, the definition of output and the multiplicity of factors that may influence the learning process. For instance, relevant factors like some teacher characteristics, the innate capacities of students and the interaction with colleagues are difficult to incorporate into an empirical model. Additionally, the relationship between inputs and output in the educational process is rather complex and can only be summarised imperfectly in a production function. Such difficulties have been addressed in detail in the education economics literature and we will touch upon them in the course of this article. Production frontier estimation in the field of education has mainly used non-parametric techniques like the Data Envelopment Analysis and the Free Disposable Hull (FDH), sometimes complemented with regression analysis (see, for instance, Bessent et al. (1982), Ray (1991) and Ruggiero (1996)). SFA has been already used in this context as well, like in Mizala et al. (2002). This methodology is more de manding in terms of assumptions, since it requires the specification of a functional form for the production function, but it is less sensitive to the presence of outliers and allows the possibility of making inference about the contribution of inputs.