Forecasting Portuguese GDP with factor models: Pre- and post-crisis evidence

Abstract

In this article, we assess the relative performance of factor models to forecast GDP growth in Portugal. A large dataset is compiled for the Portuguese economy and its usefulness for nowcasting and short-term forecasting is investigated. Since, in practice, one has to cope with different publication lags and unbalanced data, we also address the pseudo real-time performance of such models. Furthermore, by considering a relatively long out-of-sample period, we are able to evaluate the behavior of the different models over the pre-crisis period and during the latest economic and financial crisis. As Portugal was one of the hardest hit economies, it is a particularly insightful case to assess the relative performance of factor models during a period of economic stress.