Year of Graduation
Forecasting Of Key Macroeconomic Variables With Frequentist And Bayesian Methods
The aim of this work is to build forecasts for three macroindicators of the Russian economy: CPI index, indices of industrial production and bank rate using BVAR models. The peculiarity of the work is the determination of the optimal parameter and lag length: the method of maximizing marginal density was used, which was never used to predict Russian data.Also, several prior distributions are used: the Minnesota distribution, the conjugate normal-inverse Wishart prior distribution, and the Diffuse Jeffreys distribution (first applied to the construction of forecasts of Russian macro variables). As a result, the model with prior distribution of Minnesota and the fixed parameter value is optimal for all variables.