Прогнозування розвитку економіки України на основі баєсівських авторегресійних (BVAR) моделей з різними priors

In this article the theoretical analysis and practical application of Bayesian approach for vector autoregressive model parameters estimation with different priors have been peformed. The time series was from 2001Q1 to 2010Q4 and included the following variables: GDP, CPI, exchange rate, unemployment level, nominal long-term interest rate, and gas and oil prices. Comparative analysis of nineteen received models showed, that the better results were received in the frames of BVAR(2) model with Minnesota priors. Based on this model, the forecast and impulse responses on 24 quarter ahead time horizon were also done.