We employ principal components and dynamic factor models for nowcasting GDP growth in selected Central, Eastern and Southeastern European (CESEE) economies. Our estimation sample extends from the first quarter of 2000 to the second quarter of 2008, our evaluation period from the third quarter of 2008 to the third quarter of 2014. For this period, we produce quasi out-of-sample forecasts of past-, current- and next-quarter GDP growth for seven CESEE economies. The models differ with respect to the estimation method, model specification, and the number of short-term indicators used. We find, first of all, a clear gain in predictive accuracy from using a nowcasting model with monthly indicators compared to the naïve benchmark. Furthermore, for our sample of small, open economies, we find that models using a smaller set of carefully selected indicators yield lower prediction errors on average than models based on larger information sets. Finally, we identify a clear gain in forecast performance from including foreign or euro area indicators.
Authors: Havrlant, D., Tóth, P., Wörz, J.
