Um modelo para projeção de arrecadação tributária dos entes subnacionais

The forecasting of government revenues is extremely important for an adequate budget execution, since a good accuracy in the estimation allows the stipulation of an expenditure level that meets the demands of the population. Using data released by the National Council of Treasury Policy, the general objective of this study is to make monthly projections of disaggregated state tax revenue using the dynamic factorial model. These projections are conditioned to the official medium-term macroeconomic scenarios related to prices and economic activity. The results obtained were reasonable, and the accumulated percentage error for six months was around 10%. This considering a sample composed of more than 400 series containing quite disparate trend and seasonality patterns, besides the problem of structural breaks, measurement error, missing data, etc. This study presents a proposal for state tax revenue projections that will benefit public managers in fiscal planning.