Small-scale nowcasting models of GDP for selected CESEE countries

In this article, we describe short-term forecasting models of economic activity for seven countries in Central, Eastern and Southeastern Europe (CESEE) and compare their forecasting performance since the outbreak of the Great Recession. To build these models, we use four variants of bridge equations and a dynamic factor model for each country. Given the differences in availability of monthly indicators across countries and the rather short time period over which these indicators are available, we favor smallscale forecasting models. We selected monthly indicators on the basis of expert judgment, correlation analysis and Bayesian model averaging techniques. While our odels generally outperform a purely time-series based forecast for all CESEE countries, there is no single technique that consistently produces the best out-of-sample forecast. To maximize forecasting accuracy, we therefore recommend selecting a country-specific suite of well-performing models for every CESEE economy.