Multiple dimensions of regional economic growth: The Brazilian case, 1991-20001

This paper seeks to understand how and why the determinants of economic growth (including spatial spillovers) in Brazil may manifest themselves differently at different spatial scales (municipalities, micro-regions, spatial clusters, and states) between 1991 and 2000. Analysing this issue it sheds light on the geography of the structural process underlying the economic growth at different scales. It means that the definition of each scale level could have a well-defined role in the economic growth process. A complementary approach is related to the Modifiable Areal Unit Problem (MAUP) and Ecological Fallacy Problem. These two measurement problems stem from the fact that there is an aggregation problem which might prevent us from identifying the actual scale at which processes operate. This paper suggests a general framework that allows dealing with multiple spatial scales, spatial autocorrelation and model uncertainty. The analysis reveals that if single regression is estimated at the different scale levels, the results change as scale level changes. However, the robustness test was able to identify variables that are simultaneously significant at different spatial scales: higher education and health capital and better local infra-structure are related to higher economic growth rates. Among other results, this paper identifies that spatial spillovers are operating especially at finer scales. At municipal level, several variables exhibit externality effects across spac