No abstract

This paper combines classic and spatial shift-share decompositions of 1981 to 2006 employment change across the 27 states of Brazil. The classic shift-share method shows higher employment growth rates for underdeveloped regions that are due to an advantageous industry-mix and also due to additional job creation, commonly referred to as the competitive effect. Alternative decompositions proposed in the literature do not change this broad conclusion. Further examination employing exploratory spatial data analysis (ESDA) shows spatial correlation of both the industry-mix and the competitive effects. Considering that until the 1960s economic activities were more concentrated in southern regions of Brazil than they are nowadays, these results support beta convergence theories but also find evidence of agglomeration effects. Additionally, a very simple spatial decomposition is proposed that accounts for the spatially-weighted growth of surrounding states. Favourable growth in northern and centre-western states is basically associated with those states' strengths in potential spatial spillover effect and in spatial competitive effect.

No abstract

In this paper, we study the hitherto unexplored evolution of the size distribution of 185 urban areas in Brazil between 1907 and 2008. We find that the power law parameter of the size distribution of the 100 largest urban areas increases from 0.63 in 1907 to 0.89 in 2008, which confirms an agglomeration process in which the size distribution has become more unequal. A panel fixed effects model pooling the same range of urban size distributions provides a power law parameter equal to 0.53, smaller than those from cross-sectional estimation. Clearly, Zipf's Law is rejected. The lognormal distribution fits the city size distribution quite well until the 1940s, but since then applies to small and medium size cities only. These results are consistent with our understanding of historical-political and socio-economic processes that have shaped the development of Brazilian cities.