Nowcasting GDP in Greece: A Note on Forecasting Improvements from the Use of Bridge Models

In the recent literature on nowcasting, the use of the so-called bridge models is advocated. These are simple regression models that use data on mixed frequencies, usually with the lower frequency data serving as dependent variables and the higher frequency data as explanatory variables. In this note we investigate whether the use of such models can lead to performance enhancements in forecasting real GDP growth for Greece. This is an interesting and instructive exercise because of the obvious break in Greek real GDP growth during the crisis but also, and more importantly, because of the potential usefulness of such models in forecasting the anticipated recovery in Greek growth. Since many monthly activity indicators are released in advance of GDP growth it is interesting to see how the structure and timing of bridge models can lead to potential improvements in forecasting growth. Our results indicate that by using three of the most important monthly activity indicators such performance enhancements are indeed possible.