Are Spectral Estimators Useful for Implementing Long-Run Restrictions in SVARs?

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, thiscould 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.