Answers are grounded in this paper and cite their sources.
This paper asks whether FOMC minutes improve forecasts of future monetary policy beyond the information already reflected in market prices. We study two forecasting tasks. First, we test whether federal funds futures rates become more accurate after minutes publication. For each release, we compare the same contract ten minutes before and twenty minutes after publication with the same realized monthly average effective federal funds rate. Matched non-release days distinguish unusually large price revisions from improvements in forecast accuracy. A forecast-revision identity separates revision size from alignment with the eventual outcome and provides a conditional-mean benchmark. This benchmark explains why a large price response need not improve a forecast: the revision must be sufficiently aligned with the future realization to offset its squared magnitude. Second, we test whether the language of the minutes improves predictions of the next policy decision: a cut, hold, or hike. A supervised sentence classifier assigns hawkish, dovish, and neutral probabilities. Hawkish language favors tighter monetary policy, while dovish language favors easier policy. Averaging the hawkish-minus-dovish probability yields a document-level tone score. We add this score and related text measures to models using Treasury yields and policy history, refitting chronologically with verified source dates. This ordering prevents later source documents from entering earlier measurement fits, while retaining the retrospective nature of the analysis. Across 104 releases in 2010–2024, the change in next-month mean squared forecast error is +0.155 squared basis points (95% block-bootstrap interval [−0.530, 1.069]). Minutes-release windows show larger revisions than matched controls at horizons of one to six months, but no clear forecast-accuracy advantage. Across 160 meetings in 2003–2022, the text measures do not reduce forecast log loss relative to the stronger Treasury benchmarks. The evidence distinguishes a market response to minutes from an incremental forecasting gain, which is not established in these retrospective tests.
No takes yet. Share an insight, caveat, or question.
Miquel Noguer Alonso (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: