The main goal of this paper is to determine which variables have a robust correlationwith the growth rate of per capita Gross Domestic Product (GDP) of Brazilianstates between 1960 and 2000. We have run two tests of robustness suggested bythe literature. The first approach is the Extreme Bounds Analysis (EBA) testproposed by Levine & Renelt (1992). An alternative approach to the previous onewas considered by Sala-i-Martin (1997). The latter author argues that instead ofanalyzing the extremities of the coefficients estimates of a specific variable, it isnecessary to make the analysis of the distribution of all coefficients of this variable.In sum, based on those tests, we can affirm that urbanization, mortality rates,fertility rates, pluviometer, tax burden and migration have a robust correlation withthe growth rates of per capita GDP of the Brazilian states. Moreover, it was notdenied the occurrence of conditional convergence for the Brazilian states.
