Modelos univariados de séries temporais para previsões de curto prazo da arrecadação nacional do FGTS

Abstract

To formulate statistical models directed to the accomplishment of forecasts in short term (up to 12 steps to the front) on the collection of the FGTS, with sights to contribute with the precision of the estimates contained in its annual budget is the objective of the present research. The resources of the FGTS constitute one significant funding it federal government to finance workmanships of habitation, sanitation and infrastructure, generating of great economic and social impact. Caixa Econômica Federal is the operator agent of the FGTS, for that it has the responsibility to manage the collection of its contributions, under the supervision of the Advice Custodian of the FGTS - CCFGTS and the Ministry of the Cities, managing agency of the applications and responsible for the elaboration of the Annual Budget. To each end of year forecasts on the collection of the FGTS with sights are made to subsidize the budgetary and financial planning of its resources. Such forecasts have a crucial importance, in the measure where small margins of error can generate expressives distortions, considering the volume of the involved values in absolute terms, situation that demands one high degree of precision. For analysis and modeling of the series of collection of the FGTS the models of exponential smoothing and ARIMA had been used, beyond the technique of combination of forecasts. The results of the done forecasts to leave of these models had been better that the constant in the budget of the FGTS, what it suggests the necessity of revision of the techniques applied in the current model.