Análise do Índice de Preços ao Consumidor Amplo (IPCA) por meio de Combinações de Previsões
Análise do Índice de Preços ao Consumidor Amplo (IPCA) por meio de Combinações de Previsões
-
DOI: https://doi.org/10.22533/at.ed.7232430012
-
Palavras-chave: Índice de Preços ao Consumidor Amplo – IPCA; Combinação de Previsões; Modelagem
-
Keywords: Índice de Preços ao Consumidor Amplo – IPCA; Forecast Combination; Modeling
-
Abstract: The Índice de Preços ao Consumidor Amplo (IPCA) aims to measure the inflation rate based on a basket of goods and services traded at retail, representing the personal consumption of households. Staying updated on economic indicators is of utmost importance to understand the trends and the situation of the Brazilian economy. These indicators provide relevant data on inflation, the performance of the productive sector, price variations, and household income, among other essential aspects. Being informed about such metrics enables a comprehensive view of the country's economic landscape. Given the importance of this index, the objective of this study is to forecast future values of the CPI using forecast combinations. The base models used are SARIMA, Exponential Smoothing Models, and Holt-Winters, totaling six adjusted models. To obtain more accurate forecasts, forecast combination methods were applied, including simple arithmetic mean, minimum variance, and linear regressions (OLS) and robust regressions (LAD). To evaluate the accuracy of the forecasts and achieve the proposed objective, the following performance measures were used: RMSE, MAPE, and MAE. Based on the accuracy measures, three combinations were selected, for which forecasts were calculated for the year 2023: the combination using least squares and robust regression with six models, and the robust regression using the SARIMA, Additive Holt-Winters, and Multiplicative Holt-Winters models with damping factor.
- Cleber Bisognin
- Glaucio Jorge Ferreira Rosa
- Diego Brenner Dos Reis