Institutions, Geography, Regions, Countries and the Mobility Bias

Recent research on the ultimate causes of the large differences in economic development across countries has framed the issue as a competition between geography and institutions. Some papers claim to have evidence that geography explains nothing after controlling for institutions. This paper argues and shows evidence that geography correlates with economic activity across regions within countries where institutions are far more constant than across countries. The paper also argues that mobility of factors can in theory completely mask the impact of geography if one looks only at output per-capita. Comparing cross-region with cross-country data and comparing regressions of output density with regressions of output per-capita, the paper finds evidence of this mobility bias. Finally the paper tries to integrate re-cent lines of research with an earlier line that looked at regional data as a laboratory to test for economies of scale. Overall the paper finds evidence for three important determinants of spatial income levels: geography, institutions and agglomeration economies associated with very large urban areas. One needs to consider all three to fully explain the pattern of economic activity across space.