Please use this identifier to cite or link to this item: http://hdl.handle.net/10419/128052
Authors: 
Mertens, Elmar
Year of Publication: 
2008
Series/Report no.: 
Working Paper, Study Center Gerzensee 08.01
Abstract: 
No, not really. Responding to lingering concerns about the reliability of SVARs, Christiano et al (NBER Macro Annual, 2006, "CEV") propose to combine OLS estimates of a VAR with a spectral estimate of long-run variance. In principle, this could help alleviate specification problems of SVARs in identifying long-run shocks. But in practice, spectral estimators suffer from small sample biases similar to those from VARs. Moreover, the spectral estimates contain information about serial correlation in VAR residuals and the VAR dynamics must be adjusted accordingly. Otherwise, a naive application of the CEV procedure would misrepresent the data's variance.
Subjects: 
Structural VAR
Long-Run Identification
Non-parametric Estimation
Factorization of Spectral Density
JEL: 
C32
E17
Document Type: 
Working Paper

Files in This Item:
File
Size
706.25 kB





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.