A novel content-based approach to measuring monetary policy uncertainty using fine-tuned LLMs

Policy uncertainty is a potential source for reducing policy effectiveness. Existing studies have measured policy uncertainty by tracking the frequency of specific keywords in newspaper articles. However, this keyword-based approach fails to account for the context of articles and differentiate the types of uncertainty that such contexts indicate. This study introduces a new method for measuring different types of policy uncertainty in news content using large language models (LLMs). We fine-tune the LLMs to identify different types of uncertainty expressed in newspaper articles based on their context, even if they do not contain specific keywords indicating uncertainty. By applying this method to Japan’s monetary policy from 2015 to 2016, we demonstrate that our approach successfully captures the dynamics of monetary policy uncertainty, which vary significantly depending on the type of uncertainty examined.