The agglomeration effects on growth have been comprehensively documented and studied in growth literature, yet, the same cannot be said for the reverse causality, which poses endogeneity problems for the estimation of growth models. In addition, most, if not all, of the empirical research in the area use some conventional indicators such as labour density to proxy the level of agglomeration of an economy, without acknowledging the actual spatial structure of the economic activities that could be more fully captured by spatial autocorrelation (s.a.) measures. In light of this, this paper attempts to shed additional light on the growth literature by examining the Granger causality between growth and agglomeration using spatial autocorrelation techniques in a VAR framework. Our dataset covers the 26 states of Brazil for the period of 2001-2009. The results suggest that (i) Spatially autocorrelation is evident in the Brazilian economy across the 26 states; (ii) growth, proxied by GDP per capita, can be explained by s.a. of economic activity and labour density; (iii) growth is depressed by s.a. of labour distribution; (iv) growth fosters spatial clustering of economic activity; (v) growth triggers greater within-state spatial variation of GDP per head; and (vi) growth reduces labour density but induces s.a. of labour distribution.