Abstract
This paper examines whether statistical natural language processing techniques have been useful in analyzing documents on monetary policy. A simple latent semantic analysis shows a relatively good performance in classifying the Bank of Japan (BOJ)’s documents on its governors’ policy and the impact without human reading. Our results also show that Governor Haruhiko Kuroda’s communication strategy changed slightly in 2016 when the BOJ introduced the negative interest rate policy. This change in 2016 is comparable to the one from the transition from Masaaki Shirakawa to Kuroda. In spite of the intention, the BOJ had a misjudgment in the communication strategy.

Creation Date/NO. February 2017 17-E-011
Research Project Sustainable Growth and Macroeconomic Policy