by Serge Coulombe, University of Ottawa, December 2003
Abstract
This paper provides an empirical analysis of the comparative evolution of interprovincial and international trade and their effects on regional growth for the Canadian provinces since 1981. First, we establish a striking empirical fact: the 'L' curve that characterized the comparative evolution of interprovincial and international trade shares to gross domestic product (GDP) between 1981 and 2000. In the 1980s, the interprovincial trade share was falling while the international trade share was constant. A sharp break occurred around 1991 and, throughout the 1990s, the international trade share expanded rapidly while the interprovincial trade share remained constant. The analysis casts doubt on the pure diversion model often used in trade modelling, as in the structural gravity model of Anderson and van Wincoop (2003) used recently to revisit the Canada-U.S. border effect. In the second part of the paper, we use a conditional convergence-growth model to estimate the respective long-run effects of interprovincial trade and international trade on Canadian regional economies. It appears that international trade creates jobs and generates higher productivity. In contrast, interprovincial trade only creates jobs. In the long run, a 10 percent increase in international trade share translates into an increase in relative GDP per capita and in labour productivity of 6.3 and 4.0 percent, respectively. The long-run effect of a 10 percent increase in interprovincial trade on per capita GDP is 5.1 percent, but the effect on labour productivity is virtually nil.