Avaliando o forecast content e o forecast horizon de modelos ARMA para os principais agregados de arrecadação de ICMS do Estado do Ceará

• Ano de defesa: 2019
• Autor(a) principal: Matos, Mariana Camelo Sá
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Arrecadação de ICMS; Modelo ARMA; Forecast Content; Forecast Horizon; Séries Temporais; Previsão
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/50123

The ARMA model is a simple tool which is widely used in research for predicting future values of time series. For this reason, this dissertation's main goal is to evaluate the efficiency of Forecast Content and Forecast Horizon of ARMA models of total tax revenue from ICMS in Ceara. In other words, we investigate how well ARMA model preserves its ability to predict ICMS better than average as we increase the time horizon. Our dataset was extracted from Boletim de Arrecadacao Mensal dos Estados e do Distrito Federal between January 1997 and May 2019, available at CONFAZ's official website. Our results show that ARMA keeps good performance in a horizon of 6-12 months. This should be taken as a reference (or standard model) when predicting the state's tax revenue.