A gravity model for estimating passenger origin-destination flows between countries worldwide

Abstract

The purpose of this paper is to present a gravity model for estimating passenger origin-destination (OD) flows between countries worldwide. These model results can subsequently serve as input to identify airport-airport OD passenger flows. Based on data from Sabre MI and the World Development Indicators we compare gravity models estimated in a log-linearized form by traditional Ordinary Least Squares (OLS) and in a multiplicative form by Poisson pseudo-maximum-likelihood (PPML). The dependent variable to be modelled is the number of OD passengers per year between two countries. Independent variables include factors such as GDP per capita, population and air fares. We expect to find a significant relationship between the annual OD passengers between two countries and the explanatory variables. While distance- and GDP-related variables are rather common in gravity models, we expect further insight by inclusion of the air fare variable, which has been computed on basis of Sabre MI data. Furthermore, we expect the PPML estimator to produce better estimation results than OLS. The main contribution of this paper is a global OD passenger model on country level. We have included the fare variable to gain further insight into the structure of global passenger OD flows between countries. Furthermore, we have employed a PPML estimator to produce better and more reliable forecast results.