DC Accepted Papers Paper:
How to Model Forward Guidance and Address a Larger Puzzle
*Names in bold indicate Presenter
Forward guidance is commonly modeled as a single anticipated future policy shock. I, instead, model forward guidance as a sequence of anticipated future policy shocks, calibrated to the high-frequency movement of the expected path of the policy rate around monetary policy announcements in the U.S. during the zero lower bound period. This, in turn, adds to the "forward guidance puzzle" in a standard New Keynesian model. I estimate structural parameters from various models of expectation formation to match the empirical responses of macroeconomic variables to a forward guidance surprise. I show that the class of models that introduces a discount parameter to address the puzzle fails to imply empirically plausible structural parameters. For example, the behavioral New Keynesian model of Gabaix (2016) requires an implausibly small cognitive discount parameter. Instead, I illustrate that the sticky information model of Reis (2009) addresses the puzzle with an empirically plausible rational inattention parameter. Lastly, I show that introducing uncertainty about the nature of the forward guidance shock in a standard New Keynesian model also addresses the puzzle with a modest probability of Delphic interpretation.
