Deciphering Federal Reserve Communication via Text Analysis of Alternative FOMC Statements

Thu, 01 Jan 2026 00:00:00 +0000

This paper proposes a text-based measure of monetary policy stance by modelling FOMC post-meeting statements as convex combinations of the staff-drafted dovish (“alternative A”) and hawkish (“alternative C/D”) versions that accompany each meeting, providing a transparent and adaptive reference spectrum. The authors fine-tune the Universal Sentence Encoder—a pre-trained language model—using synthetic examples that mirror numerical information in policy actions, enabling the model to capture both semantic tone and quantitative precision. Stance is defined as the product of tone (alignment with the dovish/hawkish alternatives) and novelty (semantic shift from the previous statement), and is decomposed into expected and surprise components using intraday financial data. Surprises arise from shifts in tone relative to market expectations or from statement novelty. The resulting surprise measure aligns closely (correlations of 70–80%) with established high-frequency measures (Swanson 2017, Nakamura-Steinsson 2018, Bauer-Swanson 2023), and the framework enables counterfactual analysis of how alternative communication could have moved markets.

Summary of a forthcoming paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.

In depth

Q1. What are the alternative FOMC statements and how are they used?

For each FOMC meeting, staff draft multiple versions of the policy statement—typically a more dovish “alternative A,” a baseline “alternative B,” and a more hawkish “alternative C” or “D”—and the paper uses these pre-structured alternatives as a reference spectrum against which to position the released statement. This institutional feature provides a transparent, adaptive measure of tone that evolves with the policy environment and internal deliberations, avoiding the rigidity of pre-fixed tone definitions. The released statement’s embedding in the language model space is compared to the dovish and hawkish alternatives to determine its location on the policy spectrum.

Q2. How is the policy stance measure constructed?

Stance is defined as the product of novelty and tone: novelty captures semantic shifts from previous statements, and tone reflects the alignment of the released statement with the dovish or hawkish alternatives; the observed stance reflects both the content of each position and the relative positioning of the Committee along the policy spectrum. A second, structural interpretation models the released statement as the outcome of internal deliberation—a weighted average of dovish and hawkish stances—linking textual variation to shifts in the internal balance of influence within the Committee.

Q3. How is the stance decomposed into expected and surprise components?

The decomposition into expected and surprise components uses intraday bond price movements to recover the market-expected dovish weight of the released statement, then defines the surprise as the deviation between the realized stance and the market-expected stance. Surprises arise from two sources: deviations in tone relative to expectations, and statement novelty. This framework shows that monetary policy surprises are not just about what the Fed did but also about how it communicated—capturing interpretable surprises that reveal shifts in the Committee’s internal balance.

Q4. How is the measure validated and what are its macroeconomic effects?

The surprise measure aligns closely with established high-frequency measures (correlations of 70–80% with Swanson 2017, Nakamura-Steinsson 2018, and Bauer-Swanson 2023); surprise tightenings reduce stock prices, raise short-term Treasury yields, dampen real activity and inflation, and raise credit risk premia. Local projection estimates corroborate that surprise contractionary shocks have the expected macroeconomic effects, providing a validation that the text-based measure captures meaningful monetary policy information beyond what is already priced in.

Q5. What counterfactual analysis does the framework enable?

The framework enables counterfactual analysis of how alternative FOMC communication could have moved markets—for example, estimating what asset price movements would have occurred had the Committee released the more dovish or hawkish alternative statement rather than the actual release. This counterfactual capability stems from the explicit modelling of stance as a position on a spectrum defined by the staff-drafted alternatives, so the market impact of any point on that spectrum can be estimated.

Key concepts

alternative FOMC statements : staff-drafted dovish (“alternative A”) and hawkish (“alternative C/D”) versions of the FOMC post-meeting statement prepared for each meeting; used as the reference spectrum for measuring the tone and position of the released statement. monetary policy stance : as defined in this paper, the product of tone (alignment with the dovish/hawkish alternatives) and novelty (semantic shift from the previous statement); captures both the direction and the information content of the released statement. tone : the alignment of a released FOMC statement with the dovish or hawkish alternative drafts in the Universal Sentence Encoder embedding space; reflects the direction of the Committee’s communication along the policy spectrum. novelty : the semantic distance of the released FOMC statement from the previous statement in the embedding space; captures how much new information or emphasis the statement introduces. Universal Sentence Encoder (USE) : the pre-trained language model applied by the paper; fine-tuned on synthetic examples that mirror numerical information in policy actions (e.g., rate-hike sizes) to capture both semantic tone and quantitative policy precision.

Text-Analysis | Macro Paper Warehouse