Recent research shows that when commonly estimated dynamic Taylor rules, which are augmented with a lagged interest, are embedded in a variety of macroeconomic models, they imply a greater amount of predictable information about future movements in interest rates than is actually evident in the yield curve. We extend the analysis to consider more generally the predictability of the arguments of the Taylor rule—inflation and the output gap—in addition to the interest rate. Specifically, we compare the predictability of these three variables in a macroeconomic model with a dynamic Taylor rule to their predictability in real-time surveys of macroeconomic forecasters or a VAR model. We find that the strongest evidence against the dynamic Taylor rule is that while it is easy to predict the variables that enter the rule, it is very hard to predict actual interest rate changes. This disparity suggests that dynamic Taylor rules neglect important aspects of monetary policy behavior.