Bayesian approach to forecasting macroeconomic aggregates

The aim of this thesis is the introduction to Bayesian statistics and econometrics, summary of the historical development of this approach and to highlight the possibility of using Bayesian approach in macroeconomics, where Bayesian approach is one of the possibilities to forecast macroeconomic aggregates such as gross domestic product, inflation, exchange rate etc. Bayesian methods are currently increasingly used due to their indisputable advantages. Macroeconometric Bayesian world is often forced to work with the multivariate autoregressive models in which the analytical expression of the marginal posterior densities of parameters of the individual explanatory variables and their lags is almost impossible. Therefore Bayesian econometrics use MCMC methods that are able to simulate these posterior densities. This thesis deals with description and application of Bayesian approach to estimate the unknown parameters of the autoregressive models and the following application to real data of the Czech Republic. The simulations of the posterior distributions of the autoregressive parameters are carried out with the help of one of the most frequently used MCMC method called Gibbs sampler.