Previsão da produção industrial do Amazonas e do estado de São Paulo através de sua produção industrial agregada ou desagregada, ou a combinação delas

Resumo

This work investigates whether the use of an aggregated or disaggregated series or a combination of both improves the prediction of the aggregate series. Econometric techniques, such as the selection operator and the Autometrics algorithm, are used to predict the first difference in the logarithm of industrial production in the states of Amazonas and São Paulo from January 2002 to December 2023. The disaggregated series showed better performance in forecasts from two to twelve months for São Paulo, using the mean square error (MSE). For Amazonas, the disaggregated series had a better performance in horizons of one to four months, while the aggregate series was higher for horizons of five to twelve months. The final test indicated that, in Amazonas, the Autometrics model with the disaggregated series was the most efficient for horizon of one month. In São Paulo, the combination of AR(13) and Autometrics models was more efficient in different horizons: aggregated and disaggregated series for two months, aggregated for nine and twelve months, and disaggregated for eleven months.

Ficha do documento

Tipo

Dissertação

Ano

2024

Instituição

Fundação Getulio Vargas

Fonte

Repositório da FGV

Idioma

Português

Acesso

Acesso aberto

Identificador

oai:repositorio.fgv.br:10438/35833

Temas

EconomiaDados

Palavras-chave

Produção industrialPrevisãoSeleção de modeloIndustrial productionForecastModel selectionEconomiaProdução (Teoria econômica)Previsão econômicaIndicadores econômicosModelos econométricos