Do measures of monetary policy in a VAR make sense?

No. In many vector autoregressions (VARs), monetary policy shocks are identified with the least squares residuals from a regression of the federal funds rate on an assortment of variables. Such regressions appear to be structurally fragile and are at odds with other evidence on the nature of the U.S. Federal Reserve's reaction function; furthermore, the residuals from these regressions show little correlation across various VARs or with funds rate shocks that are derived from forward-looking financial markets. My results provide a sharp critique of current monetary VARs.