← Back
Evidence source 5727Spot Checked

Monetary Policy Schocks: A New Hope Large Language Models and Central Bank Communication

BAFFI CAREFIN Centre Research Paper2025-10-01Paper
Executive summary

Researchers have developed a new method to measure monetary policy surprises using a multi-agent Large Language Model system to analyze the Federal Reserve’s public documents, such as Beige Books and Federal Open Market Committee (FOMC) Minutes. This approach explains over 50 percent of policy rate changes, far surpassing traditional market-based methods. The system offers real-time, transparent insights and enables more accurate predictions for economic and financial impacts, though it may miss private information available only to the Fed.

What it examines

This paper introduces a new method using large language models (LLMs) to analyze Federal Reserve communications and measure unexpected monetary policy decisions. By processing documents like the Beige Book and Minutes before each meeting, the approach aims to create cleaner, more accurate measures of policy surprises than traditional market-based methods.

What it concludes

The results show that LLM-based analysis provides more reliable and timely identification of monetary policy shocks, leading to better economic predictions and trading strategies. This method can help central banks, researchers, and financial professionals understand and react to policy changes, and could be adapted for other areas involving important institutional communications.

Extracted from this source

Evidence objects

Evidence 592378% extraction confidence
Researchers unveil a novel multi-agent LLM system that analyzes Federal Reserve communications, like Beige Books and FOMC Minutes, to measure monetary policy surprises more accurately than traditional market-based or narrative methods.

key_findings bullet 1 · key_findings · validation V0

Evidence 592478% extraction confidence
The LLM-based approach explains over 50% of policy rate variationfar surpassing the 15-17% explained by market-based measuresenabling real-time extraction of conditional expectations and supporting more reliable economic and financial market predictions.

key_findings bullet 2 · key_findings · validation V0

Evidence 592578% extraction confidence
While the methods transparency and scalability mark a major advance, it relies solely on public information, potentially missing private Fed insights; nonetheless, it enables profitable trading strategies and introduces new terms like 'narrative surprises.'.

key_findings bullet 3 · key_findings · validation V0

Evidence 592678% extraction confidence
This paper presents a novel multi-agent LLM framework for extracting conditional expectations from Federal Reserve communications, enabling construction of narrative monetary policy surprises. Its originality lies in combining LLMs, narrative identification, and real-time analysis, yielding theoretically consistent, less noisy policy shock measures with significant implications for macroeconomics, finance, and AI-driven trading strategies.

key_findings bullet 4 · key_findings · validation V0

Raw abstract and provenance

- … profitable yield curve trading strategies that outperform … without contamination in high-frequency measures. … I test this prediction using implementable yield curve trades…

Source row: 1376 · abstract type: snippet