An analysis of gravity model distance exponents

This study examines variations in the basic parameter of the gravity model: the distance or travel time exponent. In a conceptual examination of the exponent under ideal conditions, it is noted that the exponent will be low if either production or consumption of the phenomena is geographically concentrated. It is further noted that if all regions produce and consume the phenomena, this generally implies a large number of substitute regions, and therefore a large exponent. These three variables are made operational and related to a set of eighty exponents for 1967 commodity flows in the United States. A multiple regression model is derived and utilized to estimate the exponents. The latter are then used to reestimate the flows. The root mean squared errors are quite similar for the fitted and estimated exponents. Implications of the research appear to be consistent with variations of the exponent in the urban context.