Abstract
ChatGPT is a tool that has gained much attention and has accelerated the adoption of Artificial Intelligence across many applications. Using monetary policy decisions made by the Reserve Bank of Australia (RBA), we test whether ChatGPT's classifications of monetary policy announcements are consistent with market-observed characteristics, whether this has changed with the update from GPT-3.5 to GPT-4, and whether this provides additional informativeness for changes in the yield implied by interest rate futures. Our results indicate that, regardless of statement readability, ChatGPT provides a classification of hawkish / dovish tone that is consistent with “Target Surprises” and “Path Surprises” and although it seems to improve in the latest version has more difficulty with “sentiment” classification. However, this information does not appear to help to explain changes in interest rate markets on the day of the policy announcement.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
