Measuring the Effects of a Shock to Monetary Policy: A Factor-Augmented Vector Autoregression (FAVAR) Approach with Agnostic Identification

Master Thesis by  Pooyan Amir Ahmadi
August 26, 2005


In this thesis I try to measure the dynamic effects of a shock to monetary policy in a Bayesian FAVAR framework. The innovation is to combine the Bayesian FAVAR with the agnostic identification introduced by Uhlig [2005] which has not been done yet. This identification scheme provides reasonable results and furthermore the possibility to impose a broader set of sign restriction on variables, proposed by Uhlig that are consistent with the conventional wisdom. Due to the greater information set it is possible to set the sign restrictions on several prices, monetary aggregates and short term interest rates considered in the dataset. In this vein one can narrow down the space of reasonable impulse responses in order to disentangle precisely the quantitative effects induced by contractionary monetary policy. Although the agnostic identification is a ”weaker” one with respect to the structure and restrictions imposed, this identification scheme combinedwith Markov chain Monte Carlo simulation methods delivers results that appear to be reasonable for a broad set of variables and with a higher accuracy than the alternative results provided by Bernanke, Boivin and Eliasz [2005]. Combining the two methodologies hold the enticing promise to measure the effects of a shock to monetary policy very precisly when applying it to large panels of data. From the results one can conclude that the identification scheme is crucial for a succesful identification especially when the dataset considered is large. However with increasingly restrictions the results are delivered increasingly infrequent. Additionally I provide a Matlab code for the estimation procedure.