Abstract:
After the seminal paper by McCallum, various authors have estimated the effect of
regional and national borders on trade. This paper digs deeper into the matter,
estimating how the internal and external border effect is affected by the non-linear
relation between trade and distance at different spatial levels, and the econometric
procedure used to control for it. Our paper uses a novel dataset that captures intra- and
inter-national truck shipments between Spanish regions and regions in eight European
countries. To deal with this non-linearity, we use three alternative strategies—
segmented distance, piecewise regressions and semi-parametric approaches—that
achieve similar results.