Informace o publikaci
Predikční výkon BVAR a TVP-VAR modelů
Autoři | |
---|---|
Rok publikování | 2012 |
Druh | Článek ve sborníku |
Konference | Nové trendy 2012. Sborník příspěvků z mezinárodní vědecké konference |
Citace | |
Klíčová slova | BVAR model; TVP-VAR model; stochastic volatility; forecasts |
Popis | Economic forecasting is necessary for solving decision problems which are based on knowledge of future values of economic variables. Multivariate time series forecasting is the foundation of almost all macroeconomic analysis. The paper focuses on performance evaluation of prediction models with constant coefficients compared to time-varying coefficients model. Models are applied to a four-dimensional monthly time series comprising growth real GDP, inflation, interest rate and real effective exchange rate of the Czech Republic. The forecast performance is evaluated employing Root mean square forecast errors. The results show that the time-varying coefficients model achieves much better performance forecasting than the models with constant coefficients. |