Port of destination and carrier selection for fruit exports: A multi-dimensional space-time multi-nomial probit model

This papers studies the selection of carrier and destination port for Chilean fruit exporters. This double selection is analyzed as a stochastic choice process with time and space interactions. A space-time error structure was specified within a multi-nomial probit model, considering serial correlation, spatial autocorrelation, and state dependence. The modeling approach was successfully applied to the case of grape exporters from Chile to the USA. The results validated the behavioral hypothesis for this process, i.e. there is significant state dependence, serial and spatial correlation in the choice of carrier and destination port, which should be considered when discrete choice models are used to predict export flows.