POLITICIZATION IN CENTRAL BANK DECISIONS: TEXT MINING IN THE MINUTES OF THE COPOM 2002 – 2024

Abstract

More precisely, in 2021, the Central Bank (BC) gained its independence. Although these are different periods of the economy, both in the microeconomic and macroeconomic spheres, we sought to analyze the composition of the minutes from 2003 to 2024, which ranges from the government of President Lula to the independence of the Central Bank. The objective was to verify if there was any correlation between the governments and the decisions of the Central Bank. For this, Natural Language Processing (NLP) was used, a field of artificial intelligence that focuses on the interaction between computers and human language. Next, the clustering of the data extracted from the minutes was carried out. This analysis made it possible to identify linguistic and structural patterns that could indicate changes in the BC's approach over the years, influenced by the guidelines of each government. The study aimed to explore how changes in government may have influenced the decisions of the Central Bank. The relationship of the words present in the COPOM minutes, the similarity between the minutes of different governments and the relationship of the grammatical classes used in COPOM decisions with the behavior of the basic interest rate were analyzed.