1st Annual CIRANO Workshop on Networks in Trade and Finance edition:1 location:Montreal - Canada date:09-10 November 2012
This paper extends the empirical gravity literature by including, and focusing on, the impact of networks on trade. Traditional gravity equations model bilateral trade flows as a function of income and a vector of distance variables. While the explanatory power of these models has proven to be very high, all measures of distance captured in these models are prone to some ‘ad hoc-ness’ that is related to international networks, but fail to capture more specific network elements. At the same time, the literature has mainly focused on bilateral effects between trading partners, neglecting the multilateral nature of trade in general. Network analysis presents the opportunity to address both these issues. First, network analysis allows to naturally include multilateral measures such as the number and intensity of trading partners, the effect of social networks in trade etc. Secondly, this paper estimates different weighted network measures and adds those to the vector of distance measures of the gravity models, introducing a more natural way to incorporate multilateral resistance into the bilateral models. Estimation techniques discussed are Poisson Pseudo Maximum Likelihood (PPML), Negative Binomial (NB) and zero-inflated versions of both models.