VoxEU Column Monetary Policy

Computer-based narrative analysis of the Bank of Japan’s Governor Kuroda

Communication strategies have become a policy instrument used by central banks to control expectations. This column uses a natural language processing method to explore the Bank of Japan’s communication strategy from July 2012 to November 2016, a period during which both Masaaki Shirakawa and Haruhiko Kuroda held office. The analysis suggests that since 2016, when the Bank introduced a negative interest rate policy, Kuroda's communication strategy has changed implicitly.

Communication with financial markets – a strategy in monetary policymaking by central bankers in the advanced economies – has gained momentum.1 Former Chairman of the Federal Reserve, Alan Greenspan, took office after the era of “monetary mystique” (Goodfriend 1986) and pursued a communication strategy in a collegial way, reflected by then Vice Chairman Alan Blinder (Blinder 1998). Financial markets at the time responded only to the Maestro’s voice of authority. The one-voice approach to communications is in contrast to a highly individualistic approach, whereby individual policy board members are allowed and encouraged to manifest their personal views, as is often the case with the Bank of England (Ehrmann and Fratzscher 2005). A communication strategy is now a policy instrument used by central banks to control expectations in a way that is consistent with the actual outcomes that monetary policy would achieve.

Central bankers have options for tactics to implement a communication strategy in two areas: channels and messages. Central banks communicate with financial markets through the channel of information flows in market microstructure that embed institutional settings for policy announcements. But however elaborate the timing with which central bankers may make their policy announcements for fear of bringing about unintended market impacts, there could be noises among financial markets if the policy changes are worded carelessly. Messages transmitted by the central bankers and broadcast by some media are received by market participants eager to evaluate the policy to make profits from their investments. In particular, in the exasperating course of unconventional monetary policy by some central banks, financial investors have become cautious about catching what follows the policy messages.

Some journalists have begun to poke fun at the communication strategy of the Bank of Japan (BOJ) Governor Haruhiko Kuroda, dubbing it ‘monetary shamanism’.2 Since 2013, when former Governor Masaaki Shirakawa retired, the BOJ has adopted an inflation-targeting commitment to achieving a rate of 2% in two years by doubling the amount of the monetary base. While the unconventional measures succeeded in achieving a rate higher than 2% temporarily from April 2014 to March 2015, with a peak of 3.4% in May 2014, the core inflation rate has since fallen below 0.5%, diving again into deflation and further away from the target.

Kuroda confessed to being inspired by the story of Peter Pan, quoting “[t]he moment you doubt whether you can fly, you cease forever to be able to do it.” To make the BOJ’s commitment credible to the public, Kuroda has taken every opportunity to reinforce this commitment to future inflation, in defiance of the actual state of deflation. The BOJ governor holds regular press conferences just after each monetary policy committee (MPC) meeting. At these press conferences, which are broadcast online in real time, the governor makes remarks in Japanese on the decision making of the MPC, the Japanese press asks questions. The wording used by the governor at the press conferences is covered by news media as such as Reuters.

How have the narratives issued by the ‘monetary shaman’ been perceived by online viewers or news readers? In a recent paper, we explore the BOJ’s communication strategy with a natural language processing method (Keida and Takeda 2017). Our paper follows the spirit of narrative economics (Shiller 2017) stating, “[t]he field of economics should be expanded to include serious quantitative study of changing popular narratives.” Browsing Kuroda’s statements in his regular press conferences, we analyse the communication strategy with a latent semantic analysis (LSA) in statistical natural language processing.3 All of the BOJ documents from the governor’s regular press conferences are in Japanese, requiring specific morphological analysis before LSA application. It is also impossible that the wordings are not corrected verbatim as the Wall Street Journal’s US Fed Statement Tracker4 analyses. Our analysis is able to overcome these problems that are specific to the BOJ statements.  

Our sample covers 55 BOJ regular press conferences from July 2012 to November 2016, a period during which both Shirakawa and Kuroda held office. Our LSA shows values of cosine similarity between each document. Table 1 indicates cosine similarities in a sample of press conference dates. Each governor appears to use serially correlated wordings, with cosine similarity higher than 0.8, but the relatively lower similarities of the documents across the terms suggest that Kuroda has followed a strategy that differs from that of Shirakawa. On 4 April 2013, Kuroda introduced the quantitative qualitative easing (QQE) monetary policy, with subsequent market impacts. He also decided to further ease monetary policy on 31 October 2014 with a slight majority in the MPC voting. While these two meetings indicate peculiarity among the documents issued by Kuroda, there appears to be quite strong similarities between other statements prior to 2016.

Table 1 

Since 2016, when the BOJ introduced a negative interest rate policy, Kuroda's communication strategy has changed implicitly. The cosine similarities are lower than before. The decrease after 2016 is comparable to that associated with the transition of governor from Shirakawa to Kuroda. In spite of the BOJ’s message stating that the policy would not change, as Kuroda repeatedly emphasised, our results suggest that the BOJ made a misjudgement in its communication strategy. A possible interpretation of the implicit change in the BOJ’s communication strategy might be its early consciousness of ‘exit’ from the ongoing unconventional monetary policy.

Editors' note: The main research on which this column is based appeared in a Discussion Paper of the Research Institute of Economy, Trade and Industry (RIETI) of Japan.


Blinder, A S (1998) Central Banking in Theory and Practice. The MIT Press Cambridge.

Boukus, E, and J V Rosenberg. (2006) “The Information Content of FOMC Minutes.''  Available at SSRN: https://ssrn.com/abstract=922312

Ehrmann, M, and M Fratzscher. (2005)  “How should central banks communicate?'' Working Paper Series 0557, European Central Bank.

Goodfriend, M (1986) “Monetary Mystique: Secrecy and Central Banking.''  Journal of Monetary Economics 17(1), pp. 63--92.

Holmes, D (2014) Economy of Words: Communicative Imperatives in Central Banks.  The University of Chicago Press, Chicago and London.

Keida, M and Y Takeda. (2016) “A Semantic Analysis of Monetary Shamanism: A case of the BOJ's Governor Haruhiko Kuroda.” RIETI Discussion Paper Series 17-E-011. 

Mazis, P G and A E Tsekrekos. (2015) “Content Analysis of the FOMC Statements-How does the Fed's Wording Affect Financial Markets & Capital Flows?'', paper presented at Twenty-Second Annual Conference of the Multinational Finance Society.

Shiller, R (2017) “Narrative Economics.” American Economic Review 107(4), pp. 967-1004.  


[1] The significance is also called the “necessity as the mother of invention” based on a comprehensive survey for central bank governors and academic specialists (Blinder et al., 2017)

[2] See “Central bank chiefs need to master the art of storytelling” by  Gillian Tett in the Financial Times, 23 August 23; and “Quiet critic of Kuroda’s ‘monetary shamanism’ turns up the volume” by Leika Kihara, Reuters, 21 October 2014. The former article relies on an anthropologist’s analysis (Holmes, 2014), citing “central bankers also try to control economic outcomes by using words, not merely to influence price and interest rate expectations but to shape the mood.”

[3] Applications to the US FOMC minutes are Boukus and Rosenberg, 2006; Mazis and Tsekrekos, 2015.

[iv] The Wall Street Journal's Fed Statement Tracker says that “The Federal Reserve releases a statement at the conclusion of each of its policy-setting meetings, outlining the central bank’s economic outlook and the actions it plans to take. Pundits and traders parse the changes between statements closely to see how policy makers’ views are evolving. Use the tool below to compare any two statements since 2007.”

4,619 Reads