Dynamické efekty měnové politiky v ČR: SVAR/FAVAR přístup

This thesis is focused on identifying dynamic effects of monetary shocks during the inflation targeting regime in the Czech Republic (CR) and its dependence on applied methodology. For this purpose it incorporates small-scale structural vector autoregressions (SVAR) and factor augmented vector autoregressions (FAVAR) proposed by Bernanke et al. (2005). In case of SVAR models the results indicate approximately one-year delayed decrease in prices and economic activity accompanied by contemporaneous real exchange rate appreciation in response to unexpected interest rate hike. This implies that the identification scheme proposed by Bjornland (2009) effectively eliminates delayed overshooting of the real exchange rate observed in studies investigating monetary policy effects in the CR so far (e.g. Babecká-Kucharčuková, 2013). FAVAR models in this thesis are based on information from almost two hundred time series, which substantially exceeds the only existing application of this methodology in CR by Borys and Horváth (2008), where just forty variables were used. FAVAR results suggest relatively robust evidence in terms of the direction of the monetary policy effect on key macroeconomic variables. However, compared to SVAR methodology the responses are more persistent and less statistically significant. The overall results therefore suggest that observations being at odds with underlying theory probably reflect incorrect model specification rather than the real behavior of the economy.