Combinación del método CART y de econometría de corte transversal para la identificación de determinantes: supervivencia de exportaciones argentinas

This paper identifies factors that explain the survival of Argentine trade flows using the combination of the Classification And Regression Tree analysis (CART) method and cross-section econometrics as the identification method. While the first determines the best predictors, the second offers a global function that links the dependent variable with those. The combination of these methods allows to reduce the number of potentially explanatory variables in problems in which there is no theoretical consensus, or to resolve situations in which there are non-linear relationships, missing data or anomalous observations. The application to the case study allows us to observe that there is a significant difference between pre-existing export flows and new flows, even though the explanatory variables of their duration turn out to be the same.