PROJEÇÃO DA TAXA DE JUROS SELIC NA ECONOMIA BRASILEIRA: UMA ANÁLISE COMPARATIVA ENTRE MODELOS DE PREVISÃO

Abstract

The assessment of expectations and the future behavior of interest rates in the economy is among the main areas of monetary economics, and has taken center stage in news and research around the world. Brazil adopts the inflation targeting regime, in which the Central Bank adjusts the nominal short-term interest rate (the Selic rate) according to the deviations of inflation from its target, the output gap and other macroeconomic variables. This work aims to find a forecast model for the basic interest rate. We used the structural model of Taylor's rule and the time series models VAR, ARDL and SARIMA. The analyzed period corresponds to monthly frequency data between 01/2002 to 06/2021. The results indicate that the model that provided the best prediction performance is the ARDL model. The dynamic forecast for up to 6 months ahead indicates a good relationship between the predicted values and the actual observed values of the Selic rate; The static forecast was very close to the effective values, and always within the 95% confidence interval. Comparing the forecast results of the final model identified with the forecast of financial institutions through the Central Bank's Focus bulletin, it was verified that the econometric model had superior performance, with 16% lower forecast error.