NOW-CASTING AND TEMPORAL DISAGGREGATION DYNAMIC FACTOR MODEL FOR BRAZILIAN QUARTERLY REAL GDP

The aim of this work is, firstly, to implement a method of disaggregation of the quarterly Brazilian GDP, estimating amonthly series at constant prices for the period 2003-2020, secondly to obtain predictions of a step forward with superiorpredictive capacity than the forecasts obtained by the indicator coincident IBC-Br. In order to achieve this goal, differentinterpolation techniques are tested, all represented in state-space models, and monthly coincident indicators are used as a wayto reach the best estimation monthly GDP possible. Based on a robust selection criteria, several models are estimated and thebest method of interpolating quarterly GDP is dynamic model with AR (1) errors, estimated by Kalman filter.