Measuring Monetary Policy Surprises Using Text Mining: The Case of Korea

Abstract

We propose a novel approach to measure monetary policy shocks using sentiment analy-\sis, which is relatively free from specification errors compared to VAR-identified shocks and allows time for a wider circle of market participants to digest information com-pared to shocks identified through intraday Fed futures data. We quantify the tones of news articles around 152 dates of Monetary Policy Board (MPB) meetings of the Bank of Korea (BOK) from 2005 to 2017 and then measure monetary policy surprises us-ing the changes of those tones following monetary policy announcements. We estimate its impact on asset prices and find that it better explains changes in long-term rates, while changes in the Bank of Korea’s base rate and VAR-identified monetary shocks are more closely associated with changes in short-term rates. Our result strongly suggests that a text mining approach to measure monetary policy surprises can be an useful complement to extract market expectations on future monetary policy.