Despesas públicas, crescimento e desigualdade: uma análise empírica dos governos subnacionais brasileiros

Abstract

Brazil is a country that adopts fiscal federalism, characterized by the decentralization of public revenues and responsibilities for the provision of services between the three levels of government. As a result, policy coordination becomes strategic for national development and, particularly, for the objectives of obtaining higher levels of income and reducing inequality. The aim of this study is to contribute to the understanding of how the spending policies of Brazilian states and municipalities contribute to the economic results. For this purpose, I estimated municipal short-term and long-term fiscal multipliers, considering five functions of the government (education, health, social protection, social security and sanitation), using the GMM-System method. The data constitute a dynamic panel covering 5,541 municipalities in Brazil, between the years 2011 and 2016. I also analyzed the dynamic relationships existing between state public spending, income inequality and the GDP of the states, using four different estimates: the first covering the three variables simultaneously and the others investigating the variables two by two. In these estimations, I used the Panel Vectors Auto-regression model with data from the 27 Federative Units from 2002 to 2015. Granger causality tests and the Impulse Response Functions were also computed. The results show that municipal expenditure on education and social protection have positive and significant multiplier effects, being considerably higher in the long run (i.e., there is a cumulative effect over time). The other expenses did not present significant multiplier effects. Besides, in the second analysis, I found evidence that the GDP of Brazilian states is negatively affected by income inequality and positively affected by government expenditure. Regarding the methods used, the chosen estimation methods are more appropriate, compared to other approaches, as they consider endogeneity issues and allow dynamic analysis of the data.