Central bank communication shapes financial prices and decisions about borrowing, saving and investment beyond the latest interest rate move, a speech published Nov. 12 by the Bank for International Settlements said.
The speech, “The science and practice of monetary policy communication - the European Central Bank experience”, said messages about the economy and future policy influence expectations. Those expectations affect financing conditions across markets and the decisions of households and companies.
The speech drew on recent European Central Bank staff research into monetary policy communication and discussed how the central bank communicates in practice. Its account covered both the effects of communication and changes in the tools researchers use to study what central banks say.
Communication works partly by shaping what financial market participants expect. Central banks discuss future economic developments and the factors that will influence later policy decisions, giving their messages a role beyond explaining the most recent rate decision. Through those expectations, communication affects financial prices across different maturities and levels of risk.
Those price effects in turn influence financing conditions and choices about credit, saving and investment. The relationship also extends beyond financial market participants. Households and businesses make decisions partly on the basis of their beliefs about monetary policy, and the way a central bank communicates helps shape those beliefs.
Researchers now have more ways to assess that communication because more data, greater computing power and artificial intelligence have expanded their tools. Machine learning and large language models, artificial intelligence systems that process text in context, allow researchers to analyse large amounts of written material with greater automation.
Until recently, researchers relied on dictionaries to measure the content and tone of monetary policy statements and speeches. They classified words and expressions associated with policy that restricts or supports economic activity. They also used language associated with different assessments of inflation and the economic outlook to categorise central bank messages.
The newer methods account for the context in which words appear. That differs from relying solely on counts of individual words or lists of expressions defined in advance. The distinction concerns how researchers interpret central bank language: the newer tools can consider surrounding text when assessing a message, rather than basing the assessment only on how often particular words occur.
