Mihaela Bratu

Faculty  of Cybernetics, Statistics and Economic Informatics

Bucharest University of  Economics

Bucharest, Romania

ACCURACY ASSESSEMENT OF SHORT RUN MACROECONOMIC FORECASTS IN ROMANIA

E-mail:  mihaela_mb1@yahoo.com

ABSTRACT. Inflation rate, unemployment rate and interest rate are some of the most important indicators used at macroeconomic level. These variables present an important interest for the central banks that establish the monetary policy (inflation target), but also for the government interested in public policies. Macro-econometric modeling offers the advantage of using more models to describe the evolution of a single variable and also the advantage of predicting it. But it is important to choose the forecast with the higher degree of accuracy. Calculating some indicators of accuracy we may know the best forecast that will be used to establish the macroeconomic policies. For the interest rate and unemployment rate in Romania VAR (2) models generated more accurate forecasts than ARMA models or models with lags. For the inflation rate the model with lag, which is consistent with Granger causality, determined the most accurate forecasts. The predictions based on all these models are better than those got using smoothing exponential techniques.

Received: December, 2011

1st Revision: February, 2012

Accepted: April, 2012

JEL Classification:  E21, E27,C51, C53

Keywords: forecasts, accuracy, econometric models, smoothing exponential techniques, error.