Can Business and Social Networks Explain the Border Effect Puzzle?

McCallum (1995) shows in an influential contribution that, even when controlling for the impact of bilateral distance and region size, borders sharply reduce trade volumes between countries. We use in this paper data on bilateral trade flows between 94 French regions, for 10 industries and 2 years (1978 and 1993) to study the magnitude and variations over time of trade impediments, both distance-related and (administrative) border-related. We focus on assessing the role that business and social networks can play in shaping trade patterns and explaining the border effect puzzle. Using a structural econometric approach, we show that intra-national administrative borders significantly affect trade patterns inside France. The impact is of the same order of magnitude as in Wolf (2000) for trade inside the United States. We show that more than 60\% of these (puzzling) intra-national border effects can be explained by the composition of local labour force in terms of birth place (social networks) and by inter-plants connections (business networks). In addition, controlling for these network effects reduces the impact of transport cost on trade flows by a comparable factor. Thus, business and social networks that help to reduce informational trade barriers are shown to be strong determinants of trade patterns and to explain a large part of the border puzzle