Macroeconomic Forecasts Comparisons in Romania During the Crisis Using New Methods of Assessing the Predictions Accuracy

This research brings as a novelty in literature the use of other methods to compare the forecasts accuracy: multicriterial ranks, non-parametric tests and binary logistic regression. Starting from the a conclusion stated in literature that Dobrescu model failed to predict the actual crisis in Romania, we got less accurate predictions for unemployment rate and exchange rate on forecasting horizon 2009-2011 with respect to the forecasts provided by National Commission of Prognosis (NCP). However, the inflation rate predictions based on Dobrescu model are more accurate than those of NCP. These results were gotten using the classical U1 Theils’s statistic used in making forecasts comparisons, but also the methods proposed by us that are not mentioned in literature in this context.