Using Chat GPT to Predict Canada’s Interest Rate

Abstract

This essay presents the development and the resulting predictive power of Chat GPT based sentiment index of the Bank of Canada’s (BoC) communications. The essay’s objective is to detect and quantify any signals of Forward Guidance provided in the BoC policy announcements. This essay validates the resulting index by comparison with market expectations, and internal BoC staff forecasts. The findings include significant predictive power of index to forecast future policy rate changes. The essay underscores the challenges of accurately interpreting central bank communications and the potential of AI tools in enhancing this interpretation.