Influence of Readability and Tone of the Bank of Russia Texts on Inflation Expectations

Communication policy is one of the main tools central banks use to reduce the inflation expectations of economic agents. In this paper, a modern methodology for assessing the impact of indices of communication quality on inflation expectations is applied to the texts of the Bank of Russia's news and press releases. The quality of communication policy is assessed in a variety of ways: 1) the readability of the text as assessed using the Transformer neural network on 44 textual features, 2) the conceptual complexity index (frequency and variety of economic terms in the text), and 3) the tone index (calculated using the ChatGPT-4 neural network). Indicators of the inflation expectations of the market (implied inflation) and the population (polls ot the Public Opinion Foundation) act as dependent variables. Based on the models estimated, it is concluded that the tone of press releases and news from the Bank of Russia's website has a small but statistically significant impact on the dynamics of the market's inflation expectations (the more informed part of the population). Readability does not enhance this impact. Household inflation expectations (the less informed part of agents) are not significantly affected by tone or readability. Technically, this result is a correlation, but we admit that it may be based on a causal relationship, which can be tested in the future using experimental methods.