Forecasting Brazilian GDP Under Fiscal Foresight and Nonfundamentalness with a Noncausal Fiscal VAR

Abstract

Due to the occurrence of fiscal foresight, conventional fiscal VAR models are naturally inclined to suffer from the phenomena of nonfundamentalness and noncausality, which may imply biased estimates. These problems have been widely identified in the fiscal literature, however until recently disregarded in most of the Brazilian fiscal VAR research. In order to fill this gap, we estimate a noncausal fiscal VAR model for Brazil — which addresses the misspecification arising from these problems — and use it to forecast Brazilian GDP. The results indicate that the noncausal VAR has better forecasting performance than the conventional purely causal VAR, when considering the “typical” Brazilian fiscal VAR dataset. This suggests that fiscal expectations have a crucial role in determining the dynamics of Brazilian GDP.

Keywords: VAR, fiscal foresight, nonfundamentalness, noncausality

JEL Classification: E62, H30, C53