出版社：“Victor Slăvescu‿Centre for Financial and Monetary Research
摘要：The main aim of this research is to construct different forecasts for the weight of fiscal revenues in the GDP for Romania on short horizon (2011-2013) by using different types of econometric models. Using annual data from 1995, according to Granger causality test, there is a unidirectional relationship between weight of fiscal revenues (an indicator of fiscal pressure) and real GDP rate in first difference. 74.48% of the fiscal revenues weights is due to this variable, the influence very slowly decreasing till 72.56% at the 10th lag. In the first period, the variation in transformed GDP rate explains 19.25% of the variation in fiscal pressure indicator. The predictions based on a vector-autoregressive model of order 1 (VAR(1)) outperformed the forecasts based on a Bayesian VAR model, moving average process (MA(2)) and dynamic factor model. The static and stochastic simulations based on VAR(1) generated the best predictions of the fiscal pressure indicator on the horizon 2011-2013, according to absolute and relative accuracy measures, excepting the mean error. In terms of sign and directional accuracy, all the types of forecasts performed the same.
关键词：forecasts accuracy; fiscal revenues; VAR model; impulse-response function; forecast error
