Avaliação do desempenho preditivo de modelos autoregressivos na arrecadação do IPVA de veículos novos

• Ano de defesa: 2011
• Autor(a) principal: Lopes, Paulo Sérgio Barroso
• Orientador(a): Linhares, Fabrício Carneiro
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Imposto sobre a Propriedade de Veiculos Automotores; Arrecadação
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/6241

The Tax on Motor Vehicles (property taxes) in the last ten years has shown significant growth, driven by economic and fiscal policies that encouraged increased consumption of vehicles. The property taxes rank second in volume of tax collections for the state, has an important role in the funding of municipal and state machine, the reason that led the preparation of this work. The government to dispose of forecasting models can evaluate the behavior of this important tax revenue as a way to identify which model that best suits in anticipation of this recipe. This study is a pioneer in Brazil, as evaluates the predictive performance of collection of property taxes from new vehicle, with the use of autoregressive models, with and without seasonal components. The records of the collection of property taxes from new vehicle has been taken from System property taxes, the Finance Secretary of the State of Ceará, and set the period from January 1999 to March 2010, a total of 135 (one hundred thirty-five) months. The models are evaluated on the bases for forecasts in periods of high inflow (January to March) and the low period of collection (May-July). The series was deflated by the INPC-CE to ideentify the real changes positive or negative, without inflation. Seasonality is a major feature in the series studied. The performance of the forecasts from combinations of these models will also be evaluated. The model was composed of six (6) models, plus variables for trend, seasonal dummies and component SAR (12), 3 (three) other models combined.