Mihaela Simionescu- Predictions’ Aggregation by Country to Improve the Accuracy of European Union GDP Rate Forecasts ?

The GDP forecasting is a prior concern for each country, but also for an entire region composed by more countries with specific and different evolutions of GDP. The GDP predictions for the 27 European Union are based on two econometric techniques. It was proved, using accuracy indicators like U1 and U2 Theil’s coefficients and statistical tests (Morgan-Granger-Newbold test, Harvey-Leybourne-Newbold test and Diebold-Mariano test) that the aggregation of forecasts made for each country in the EU on horizon 2011-2013 provided better forecasts than the use of a single semi-logarithmic model for the entire region. However, the naïve forecasts gave more accurate results than the proposed models, this conclusion being in accordance with the recent results in literature regarding GDP forecasting. For 2013 the ex-ante evaluation of forecasts was made, considering that the actual value is the one registered in 2012. It is anticipated an underestimation of the GDP rate according to the aggregation technique if the benchmark is the indicator’s value in 2012.