Demystifying monetary policy surprises: Fed response to financial conditions and wait and see for new economic data

This 2026 Journal of Macroeconomics paper by Zhengyang Chen asks why high-frequency monetary policy surprises (MPS) – changes in short-term interest-rate futures measured in narrow windows around FOMC announcements, which are designed to be exogenous policy shocks – are nonetheless partially and systematically predictable from information available before the meeting. Chen’s proposed explanation is that the Fed responds primarily to broad financial conditions, using them as a summary statistic for the economic outlook, while adopting a “wait-and-see” posture toward newly released economic data in the weeks just before a meeting, and that markets fail to fully account for this reaction function when they set pre-meeting expectations. The paper combines three components: a Taylor-type theoretical model in which the policy rate responds to pre-announcement financial conditions rather than directly to economic data; high-frequency event-study regressions with HAC-robust standard errors, estimated over samples of up to 169 scheduled and unscheduled FOMC announcements between January 2000 and December 2019 (the COVID period is excluded, though the authors report the pattern is robust to including it in an unreported test); and a monthly proxy SVAR estimated in levels with 12 lags over January 1995-December 2023 (with the instrument-based identification itself restricted to January 2000-December 2019) that uses the Nakamura-Steinsson (2018) surprise as an external instrument following the Gertler-Karadi (2015) approach. Using the OFR Financial Stress Index (FSI) as the main financial-conditions proxy, the paper finds that the FSI level on the day before a meeting significantly and negatively predicts several raw (unorthogonalized) MPS measures – for example a coefficient of -0.25 (t = -3.18) for the combined MPS measure and -0.31 (t = -6.06) for the Nakamura-Steinsson measure – while the FSI’s own change is insignificant on the announcement day itself but strongly positive the day after (0.37, t = 4.75), which the authors interpret as evidence that FSI responds to the monetary surprise rather than a time-varying risk premium driving the correlation. After controlling for financial conditions, a more positive real-activity surprise (Scotti 2016) predicts a more dovish MPS (coefficient -0.19, t = -3.35 for MPS), the opposite of what a Fed reacting quickly to incoming news would imply, and this “wait-and-see” predictability pattern is detectable for real-activity data released up to roughly two weeks before a meeting but fades for older data. Purging the six Bauer-Swanson (2023b) documented MPS predictors of their financial-conditions component collapses their explanatory power for MPS (adjusted R-squared falling from about 12% to roughly 3% or less across several MPS measures), and instrumenting the proxy SVAR with the Nakamura-Steinsson surprise purged of its financial-conditions correlation removes the price- and output-puzzle impulse responses that appear when the raw surprise is used as the instrument.

Journal of Macroeconomics | Macro Paper Warehouse

Demystifying monetary policy surprises: Fed response to financial conditions and wait and see for new economic data