Description

This paper contributes to the GDP revision literature by developing a novel methodology to further analyse the statistical properties of GDP series. We undertake two specific analyses: (i) To identify which of the main GDP components contributes the most to the presence (or absence) of a GDP growth revision bias, and (ii) To test the presence of a significant compositional change between data vintages. We apply these methodologies to the Chilean data for the period 2006 2019. Our main results show that the absence of a GDP growth revision bias is due to compensating effects between its main components. Finally, we don't find significant evidence for a potential compositional change among different data vintages.