The dynamics of firms’ export portfolio: A network analysis with an application to Mexico and Bulgaria

In this paper we use network-analysis tools to identify communities in the web of Bulgarian and Mexican exporters' destinations. We show that the network-based community measure is superior to extended gravity measures to predict the dynamics of firm's export-portfolio. This superiority stems from the fact that community is a revealed measure, is country-specific and can be calculated at the industry level. We show that a one unit increase in the number of destinations previously served by the exporter within a community rises between 49% and 66% the probability of selecting a new export destination in that community. The networkbased community variable improves the accuracy of the model between 8% and 17% relative to a model that only includes gravity and extended-gravity variables. We also show that industry-specific communities and general communities play similar roles in determining the dynamics of Bulgarian and Mexican exporters' portfolio.