Measuring the Impact of Monetary Policy Uncertainty from News
Monetary policy uncertainty has a quantifiable impact on equity markets and long-term interest rates all over the world. This paper features a new language model (The Narrative Monetary Policy Uncertainty Index or NMPU) that attempts to quantify the level of uncertainty from news articles more accurately than previous models that rely solely on keywords.
When the authors ran the NMPU model on a series of articles from the New York Times and Wall Street Journal from 1984-2023, they were able to more accurately predict the effect of monetary policy uncertainty on U.S. equity risk premiums than with other models. This suggests that the NMPU may be a better way to measure the level of uncertainty in news articles, and to predict the effects of that uncertainty.
The NMPU differs from prior models in that it takes a more nuanced approach to assessing uncertainty in language. For example, whereas most models are keyword-based and would flag a word like “expected” as an indicator of uncertainty in an article, the NMPU seeks additional context, and in the case of a phrase like “The economy added more jobs than expected”, the uncertainty is correctly deemed to be in the past, and does not affect the uncertainty score.
The NMPU was put to the test in a very narrow context, looking for monetary policy uncertainty specifically related to the U.S. Federal Reserve. The contextual nature of the model holds promise in improving the measurement of uncertainty related to other central banks, and to uncertainty related to oil price, inflation and more.