美联储不说话时利率意外（英）

Silvia Miranda-Agrippino|John C. Williams Interest Rate SurprisesWhen the Fed Doesn’t SpeakSilvia Miranda-AgrippinoandJohn C. WilliamsFederal Reserve Bank of New York Staff Reports, no.1178February 2026https://doi.org/10.59576/sr.1178 Abstract The predictability of monetary policy surprises based on past, public informationhas been interpreted intwo related yet fundamentally different ways. The“Fed information effect” posits that it arises due tomarkets updating their viewof the economy, based on signals implicitly revealed by the FOMC. The“Fed reactionto news” explanation posits that markets update their view of the FOMC’sreaction functioninstead. We show that interest rate surprises calculated aroundmacroeconomic releases exhibit thesamepredictability pattern as monetary policysurprises. Since these occur at a time when there is no scope formarkets to learnabout the Fed’s behavior, this pattern suggests an additional information channelunrelated to FOMC communication. JEL classification:E44, E52, E58Keywords:monetary policy surprises, Fed information effect,Fed reaction tonews,interest rate surprises,monetary policy premium This paper presents preliminary findings and is being distributed to economists and other interestedreaders solely to stimulate discussion and elicit comments. The views expressed in this paper are those ofthe author(s) and do not necessarily reflect theposition of the Federal Reserve Bank of New York or theFederal Reserve System. Any errors or omissions are the responsibility of the author(s). 1Introduction High-frequency changes in market-based interest rate expectations calculated aroundmonetary policy events—commonly known as monetary policy surprises—have becomea ubiquitous tool for the identification of the effects of monetary policy on asset prices,expectations, and the macroeconomy more broadly.1Monetary policy surprises owe muchof their popularity to the precision with which they can be calculated around the relevantpolicy announcements, leaving little if any room for other confounding events to contami-nate their signal, and therefore allowing for causal interpretation of the estimated effects.At least in principle. Users of high-frequency based identification have been confronted with two relatedissues associated with monetary policy surprises.First, the estimated responses ofmacroeconomic variables and their expectations often display signs incompatible withthe conventional understanding of the transmission mechanism of monetary policy (seeCampbell, Evans, Fisher and Justiniano, 2012; Nakamura and Steinsson, 2018; Miranda-Agrippino and Ricco, 2021, among others). Second, monetary policy surprises have beenshown to be predictable using publicly available information that predates the policyannouncements, which is inconsistent with their interpretation as “clean” measures ofmonetary policy shocks (e.g. Miranda-Agrippino, 2016; Cieslak, 2018; Bauer and Swan-son, 2023b). These patterns have been interpreted in the literature in two related yet fundamentallydifferent ways, both rooted in the premise that monetary policy surprises ought to alsocapture elements of the systematic—and hence endogenous—component of monetarypolicy for them to materialise.Proponents of the “Fed information effect” posit that they arise due to market participants also learning about the Fed’s view of the economyat the time of policy announcements, based on signals implicitly revealed by Fed officials(e.g. Nakamura and Steinsson, 2018; Jaroci´nski and Karadi, 2020; Miranda-Agrippino andRicco, 2021).2Conversely, proponents of the “Fed response to news” argument reject thenotion that the central bank has access to any different insights about the economicoutlook relative to the public, and instead argue that market participants learn aboutthe way the Fed responds to economic developments, that is, about the reaction function(Bauer and Swanson, 2023a,b). While different in interpretation, these two explanations are to a large extent obser-vationally equivalent. In particular, they both appeal to the predictability of monetarypolicy surprises, and they both give rise to responses of macroeconomic variables thatrun counter to the textbook effects of monetary policy shocks. In turn, in the empiricalidentification of monetary policy shocks, these puzzles are equivalently accounted for byprojecting monetary surprises on reduced-form expressions of the shocks to which theFed responds—e.g. as captured by the Fed’s official forecasts in Miranda-Agrippino andRicco (2021), or by macroeconomic news in Bauer and Swanson (2023a,b).3 In this paper, we propose to shed light on the relative prevalence of these two ex-planations by comparing the features—including the predictability patterns—of interestrate surprises calculated around both Fed and non-Fed events. We show that they arein many ways equivalent.In particular, interest rate surprises are predictable by pastinformation also when calculated around macr