Improving the accuracy of macroeconomic forecasts made by National Commission of Prognosis and Institute of Economic Forecasting for Romania

In this article, the accuracy of forecasts for inflation rate, unemployment, exchange rate and GDP index provided by Institute of Economic Forecasting (IEF) and National Commission of Prognosis (NCP) was assessed for the forecasting horizon 2004-2011. The hypothesis that combined forecasts is a suitable strategy of improving the predictions accuracy was tested. Only for the unemployment rate the combined forecasts based on IEF and NCP evaluations performed better than the initial forecasts. For inflation and exchange rate Dobrescu model of IEF provided better predictions, but the combined ones were more accurate than NCP expectations. The Dobrescu model predictions combined with ARMA static respectively dynamic forecasts and NCP estimations combined with ARMA static prognosis, respectively Dobrescu forecasts using EQ scheme for unemployment on a horizon of 2 years (2010-2011) improved the accuracy of forecasts made by both institutions, the combined predictions based on Dobrescu predictions and ARMA static ones using OPT scheme being the most accurate, according to U1 Theils’ statistic.