Conditional Nonparametric Frontier Models and Stylized Facts in Economics

In this paper we propose a general probabilistic formulation of the production process in which the heterogeneity of each unit is taking into account. Conditional Robust and Nonparametric Frontier Models (Daraio and Simar, 2005; 2007) are shown as a suitable tool to measure and consolidate productivity stylized facts in a rigorous way. These economic models represent the production process of decision making units in an activity analysis set up in which the performance of each unit is measured with respect to an efficient frontier estimated over the sample of observed inputs-outputs data, according to the external-environmental conditions faced by each decision mak- ing unit. The conditional nonparametric framework allows for the consideration of the idiosyncratic nature of the decision making units and open to the possibility of discover- ing new economic regularities. Moreover, in this framework, comparative institutional analysis may be shown to be both possible and measurable in practice. A general framework for derive economic regularities and let empirical evidence to contribute to the advancements of economic theory is proposed.