50th Congress of the European Regional Science Association: "Sustainable Regional Growth and Development in the Creative Knowledge Economy", 19-23 August 2010, Jönköping, Sweden
This paper investigates the contribution of regional performance to aggregate growth in the OECD. We find a great degree of heterogeneity in the performance of large (TL2) and small (TL3) OECD regions and among types (urban intermediate and rural) of regions suggesting possibilities for growth performance exist everywhere. We find that distribution in GDP and GDP per capita growth rates follow an approximately normal distribution among large regions, but the symmetry appears to break down among small regions. Distribution of each region.s contribution to aggregate growth is not symmetric. This implies that average values loose meaning and may not be the most adequate policy target for maximizing aggregate output. On the contrary, we find evidence of a power law and its scale free properties among regional contributions to aggregate growth, implying that a very small number of regions contributes disproportionately to aggregate growth whereas the most of individual regions contribute only marginally. Nevertheless, because there is a very large number of these small regions, their cumulated contribution has a major impact on aggregate growth. For the period 1995-2005, 4% of the large regions contribute to 33% of aggregate OECD growth and the remaining 96% to 67%. Confirming the scale-free properties of the distribution, only 2.4% of small regions contribute to 27% of OECD GDP growth and the remaining 97.6% to 73%. Using these properties, we then estimate an econometric model explaining growth at the regional level by several policy variables and controls, under different scenarios, and compute the derived aggregate growth. We find that a 10% increase in policy variables -- infrastructure, educational attainments, and employment growth -- increases aggregate GDP growth annually during the period 2005-2015 by 0.22 percentage points relative to the BaU scenario. An increase in policy variables by 10% in only lagging regions will entail a similar effect on aggregate output, as well as an increase in the same policies in only leading regions. The policy variable having the largest increase on aggregate growth is an improvement in human capital (measured by education attainment) at the regional level.