46th Congress of the European Regional Science Association: "Enlargement, Southern Europe and the Mediterranean", August 30th - September 3rd, 2006, Volos, Greece
The gravity model is the workhorse model to describe and explain variation in bilateral trade empirically. Consistent with both Heckscher-Ohlin models and models of imperfect competition and trade, this versatile model has proven to be very successful, explaining a large part of the variance in trade flows. However, the loglinear model cannot straightforwardly account for the occurrence of zero-valued trade flows between pairs of countries. This paper investigates the various approaches suggested to deal with zero flows. Apart from the option to omit the zero flows from the sample, various extensions of Tobit estimation, truncated regression, probit regression and substitutions for zero flows have been suggested. We argue that the choice of method should be based on both economic and econometric considerations. The sample selection model appears to fit both considerations best. Moreover, we show that the choice of method may matter greatly for the results, especially if the fraction of zero flows in the sample is large. In the end, the results surprisingly suggest that the simplest solution, to omit zero flows from the sample, often leads to acceptable results, although the sample selection model is preferred theoretically and econometrically.