什么是实证货币政策冲击？政策新闻的术语结构估计（英）

What Are EmpiricalMonetary Policy Shocks?Estimating the TermStructure of Policy News Jonathan Adams, Philip Barrett WP/25/128 IMF Working Papersdescribe research inprogress by the author(s) and are published toelicit comments and to encourage debate.The views expressed in IMF Working Papers arethose of the author(s) and do not necessarilyrepresent the views of the IMF, its Executive Board,or IMF management. 2025JUN IMF Working Paper Western HemisphereDepartment What Are Empirical Monetary Policy Shocks?Estimating the Term Structure of Policy NewsPrepared byJonathan Adams, Philip Barrett* Authorized for distribution byNigel ChalkJune 2025 IMF Working Papersdescribe research in progress by the author(s) and are published to elicitcomments and to encourage debate.The views expressed in IMF Working Papers are those of theauthor(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. ABSTRACT:Empirical monetary policy shocks (EMPS) contain information about monetary policy both todayand in the future. We define theterm structure of monetary policy newsas the marginal impact of an EMPS onthe policy residual at each horizon. Policy news at different horizons has different effects, so knowing the termstructure is necessary in order to use an EMPS to evaluate theory. We develop an IV method to estimatethisterm structure. We find that EMPS in the literature do not represent textbook policy surprises. Instead, theyrepresent a mix of information about policy at many horizons, and this mix varies depending on how the EMPSis identified. We use the estimated term structures to construct synthetic forward guidance and surprise shocks,and estimate their macroeconomic effects. Surprise interest rate hikes are contractionary with little effect onprices, while long-term forward guidance is deflationary. 1Introduction How do central bank decisions affect the economy?Empirical answers to this questionrequire well-identified monetary policy shocks.In recent years, clean identification usinghigh-frequency data (G¨urkaynak et al., 2005) or narrative methods (Romer and Romer,2004) have yielded an array of high quality empirical monetary policy shocks (EMPS). But what are these EMPS? The premise of this paper is that these approaches correctlycapture the nature of the shock – i.e. an exogenous perturbation to interest rate policy –but they may vary in their information about policy timing. That is, EMPS do not exactlyrepresent the textbook monetary policy shock, which is typically implemented immediatelyand without warning (i.e.a monetary policy surprise).Instead, each EMPS embodies adifferent combination of both policy surprisesandnews about future policies (i.e. forwardguidance shocks).1 Taking this idea seriously poses a challenge when trying to confront theory with data. Ifmodels imply that shocks with news at different horizons should have different effects, butEMPS are some combination of shocks at different horizons, then how should we interpretestimated responses to EMPS? Does the response to a given EMPS tell us something aboutthe macroeconomic effects of policy? Or just how news and surprise are combined in thatparticular empirical shock? Without a way to put some discipline on how a given EMPScombines shocks at different horizons, it is impossible to say. Our first contribution is to resolve these questions by developing a method to estimatetheterm structure of monetary policy news.This term structure decomposes an EMPS,revealing how it depends on policy surprises and news shocks for every future horizon.Our method to estimate the term structure has several stages.In the end, the estimatorhas a single closed-form expression, but it is helpful to describe it in distinct steps. First,we use plausibly exogenous macroeconomic shocks as instrumental variables in order toidentify the monetary policy rule, following insights from Barnichon and Mesters (2020).2 Next, the monetary policy residual is calculated from the estimated rule, and then strippedof correlation with lagged residuals to find the monetary policy innovations.Finally, theinnovations are regressed on lags of the EMPS to identify the term structure. It is possibleto implement these steps as a single estimator with a simple closed-form expression; weprove that it is unbiased, and derive asymptotic standard errors. Our second contribution is to estimate the term structure of sixteen well-known EMPS.We find that no EMPS is exactly a textbook monetary policy surprise.Even those withthe shortest term structures include some information several months ahead. For example,narrative-based EMPS such as that of Aruoba and Drechsel (2024) come closest to a policysurprise, yet they still contain significant information at horizons up to six months.Incontrast, EMPS intending to capture forward guidance do so effectively; they have nearlyzero news about very short-run policy. But there remains heterogeneity depending on th