Performance of Alternative BVAR Models for Forecasting Iranian Macroeconomic Variables: An Application of Gibbs Sampling

Low and stable inflation with sustainable growth is the first objective of any monetary authority. To achieve this prime goal, reliable forecast of macroeconomic variables play an important role. This paper investigates the forecasting performance of BVAR models with different priors for Iranian economy. For this purpose we use BVAR approach with Gibbs sampling for quarterly data of the Iranian economy from 1989:Q1 to 2007:Q4. The main advantage of this paper is using Gibbs Sampling to estimate BVAR models and use of Quasi BVAR models with Normal Wishart and Minnesota priors in order to compare forecast accuracy of the macroeconomic variables. Comparison of the BVAR with Gibbs Sampler and Quasi BVAR models in this experience shows that the value of MSFE in predicting macroeconomic variables for the four ahead period forecasts in BVAR model with Gibbs algorithms is less than Quasi BVAR models. Generally BVAR model with Gibbs sampling algorithms performs better than Quasi BVAR models in forecasting.