Mumbai : SBI Research, in its latest report (Issue #45, FY26), has introduced a pioneering framework using Natural Language Processing (NLP) to analyze the minutes of the Reserve Bank of India’s (RBI) Monetary Policy Committee (MPC) meetings. The report aims to quantify the "hidden meanings" and shifts in policy tone that often remain subjective in traditional analysis.
Key Highlights of the Report:
NLP Framework: For the first time, a two-layer NLP framework and an RBI-specific hawkish/dovish dictionary have been used to compute a ‘Normalized Net-Dovish Score’. This score tracks the intensity of the policy tone by calculating 'dovish minus hawkish' phrase hits per 1,000 tokens.
The Malhotra Era: The report notes that under Governor Sanjay Malhotra, the MPC is establishing a distinct narrative, currently clustered at a different intensity compared to the later tenure of former Governor Shaktikanta Das.
Tone vs. Yields: The analysis reveals that a hawkish tone does not always lead to hardening yields; even a dovish tone can sometimes result in yields hardening due to idiosyncratic market movements.
Historical Context: The report highlights that India has seen fewer rate hike episodes post the formulation of the Inflation Targeting Regime. Between 2010-2015, there were 16 rate hikes, compared to much fewer in the current regime.
Alignment of Stance: RBI's rate actions are found to be mostly aligned with its stated stance, with very few aberrations.
Dr. Soumya Kanti Ghosh, Group Chief Economic Adviser, SBI, stated that this data-driven approach allows for a clearer distinction between policy-tone intensity and broader communication style, providing valuable insights for market participants and policymakers.