GAUK - 91110 - Bayesian Econometrics and Monetary Policy
|Principal investigator:||PhDr. RNDr. Josef Stráský Ph.D.|
PhDr. Jaromír Baxa Ph.D.
Ing. Aleš Maršál M.A.
prof. Ing. Miloslav Vošvrda CSc.
|Description:||The aim of the work is using Bayesian econometrics as a tool for determination of monetary policy. Bayesian econometrics has been frequently used in models dealing with monetary policy.
Our first task is to develop a model for inflation forecasting on the basis of Bayesian vector autoregression (BVAR). Under inflation targeting, a good inflation forecasts are crucial for the conduct of monetary policy. However, traditional linear methods such as ARIMA or VAR are usually outperformed by a random walk model. On the other hand Bayesian method of estimation brings significantly better results.
Consequently, we develop a general model of monetary policy response. BVAR is the unique method that undergoes problems of overparametrization in common VAR models. Instead of strict fixation of some parameters, their values are constrained only and thus the identification remains feasible and dynamics of estimated system is sustained.
Another direction of research will demonstrate advantages of the Bayesian model averaging method. It is used under unclear choice of the independent variables. This will be connected mainly with explicit incorporation of institutions evolution into econometric models.
Interní mezi doktorandy a studenty.
|Work in grant:||Bayesovská vektorová autoregrese a další výzkum řešitelů. Prezentace na prestižních mezinárodních konferencí.|
|Finance:||290 tis. z toho
47 tis. režie
40 tis. ostatní náklady (konferenční poplatky, nákup dat)