Bagliano and Favero evaluate the VAR models commonly used to analyse the US monetary-policy transmission mechanism, focusing on three issues: specification, identification, and the consequences of omitting the long-term interest rate. They find that only VARs estimated within a single monetary-policy regime are stable and well-specified; that VAR-identified policy shocks and shocks identified from alternative sources disagree in level but agree on the qualitative transmission mechanism; and that including the long rate materially improves the estimated structure of the market for reserves and the central bank's reaction function.
"VAR-based monetary policy shocks and policy disturbances identified from alternative sources are not highly correlated but yield similar descriptions of the monetary transmission mechanism."
A careful, honest audit of the CEE-style monetary-policy-VAR programme at exactly the moment it had become the field's workhorse. The most useful finding is the identification result: that shock series from different schemes can be weakly correlated while the impulse responses they imply agree — which is the right way to be reassured, because it says the qualitative transmission story does not hinge on any one identifying assumption (a more constructive cousin of the Uhlig and Faust-Swanson-Wright critiques). The specification result is a standing caution still relevant today: monetary-VAR results are only as stable as the regime window they are estimated on. The long-rate point prefigures the later emphasis on the whole term structure in monetary transmission and is developed in the authors' companion work.