Integrating gravity: the role of scale invariance in gravity models of spatial interactions and trade

This paper revisits the ubiquitous bi-proportional gravity model and investigates the reasons why different theoretical frameworks may lead to the same empirical formula. The generic gravity equation possesses scale invariance symmetries that constrain possible theoretical explanations based on optimal allocation principles, such as neoclassical or probabilistic frameworks. These constraints imply that a representative consumer's utilities must be separable, and that an entropy model is the only consistent maximum likelihood allocation of a matrix of flows between origin and destination. The paper explores the feasibility of wider classes of non-scale invariant gravity equations, where gravity is no longer bi-proportional by including nonlinear interactions between trade costs and fundamental country factors such as economic size. It shows that such extensions are feasible but that they do not result in a significant improvement in the explanatory power of the empirical analysis.