期刊名称：International Journal of Economic Practices and Theories
印刷版ISSN：2247-7225
出版年度：2015
卷号：5
期号：2
页码：88-94
语种：English
出版社：International Journal of Economic Practices and Theories
摘要：In this paper, quarterly point forecasts and prediction intervals are built for unemployment rate in Romania. The point forecasts are based on some updated vector-autoregressive models (VAR models) and on a Bayesian VAR model. These point predictions and the root mean squared- error corresponding to the forecasts of the previous 4 quarters are used to construct the intervals. According to root mean squared-error, mean error and mean absolute error, VAR model outperformed the Bayesian approach in terms of forecast accuracy. 75% of the intervals based on VAR models included the quarterly forecasts on the horizon 2011:01-2014:04. The probability of these intervals to include the actual values is higher than 0.8, according to likelihood ratio and chi-square tests.
关键词：Econometrics and statistics;point forecasts; forecast intervals; VAR model; Bayesian VAR model;unemployment rate;C51; C53
