Nowcasting Activity in Argentina using Dynamic Factor Models

Abstract

Having a correct assessment of current business cycle conditions is one of the major challenges for monetary policy conduct. Given that GDP figures are available with a significant delay central banks are increasingly using Nowcasting as a useful tool for having an immediate perception of economic conditions. We develop a GDP growth Nowcasting exercise using a broad and restricted set of indicators to construct different models including dynamic factor models as well as a FAVAR. We compare each individual model with alternative combinations in a pseuro-real time out-ofsample exercise and find an improvement in predictive performance using the Giacomini and White (2004) test. Finally we introduce a DFM state-space approach, being able to measure the impact of data releases or news on sequential forecast.