Hawkish-Dovish Signal Analysis in Bank Indonesia's Board of Governors Meeting Press Releases Using Large Language Model

Abstract

Background: In signaling policy direction, central bank communication is crucial for aligning the expectations of financial market participants and the public with forthcoming monetary policy. This helps reduce economic uncertainty and enhance policy predictability.

Objective: This study examines the strategic role of Bank Indonesia’s communication in shaping market expectations and influencing public perception through policy direction signals in Bank Indonesia Board of Governors Meeting press releases.

Methods: Observations of these signals were conducted semi-automatically using a large language model (LLM) to analyze sentence-level interpretations by classifying each sentence according to hawkish, neutral, and dovish criteria. Using an ordered probit model, the relationship between policy direction signals in the sentences and changes in policy rates in the subsequent period was investigated.

Results: The results show that policy direction signals in press release texts are significantly positive in predicting future policy interest rates.

Conclusion: The findings confirm that hawkish and dovish signals embedded in Bank Indonesia’s monthly Board of Governors Meeting press releases significantly predict future policy interest rates, with 80% prediction accuracy. Central bank communication thus serves as a reliable monetary policy instrument that supports market expectation management, reduces financial market volatility, and enhances monetary policy credibility and effectiveness.