Práticas de política econômica: Um teste das atas do comitê de política monetária do banco central do Brasil (2003-2014)

Abstract

There is a lack of a deeper discursive analysis on the themes of economic policy and its implications, from the perspective of Political Science. The main objective of this article is to verify the viability of text analysis as a research tool on economic policy. It analyses textually the minutes of the meetings of Brazilian Central Banks's Monetary Policy Committee (COPOM) launched between 2003 and 2014, in order to identify the variation of terms and words and some positive relation among them, as well as the way economic policy was conducted along the three governments of Workers Party. This evaluation is made using Iramuteq, a software designed for textual analysis, considering some inflection moments regarding economic policy orientation, and the periods of proximity and disengagement between Central Bank and Finance Ministry. The research involves the survey of meetings minutes launched by COPOM, downloaded from Central Bank databases available on internet, its selection and analytical treatment, and the production of word clouds and similarity graphs. As hypothesis test, it expects to identify in the primary source some inflection points and terms variation. The application of text mining to economic policy studies is relatively new. The advantage of testing the hypothesis by using this tool is to try to open the ''black box'' of economic policies production. It provides, first, a better understanding of the usage of technical and economic terms and their implications; and, second, broader temporal analysis that allows us to scrutinize a huge amount of data simultaneously. It was possible to identify the variation of the used terms and its distribution throughout the time, which confirms the initial hypothesis partially. Although these variations are subtle enough, due to the treatment and the temporal restriction of the sample (it considers only governments of the same party), it is possible to claim that this type of study brings out novel analytical chances. The results point to the need to combine studies based on the use of traditional sources (official speeches and government documents) with ones that resort to primary sources (minutes), usually discharged by experts. Second, they point out there is an urgent requirement to build on the knowledge of text analysis software. Finally, they indicate there is still a great gap to be filled in the field of discursive institutionalism, especially in Brazil.