Sentiment and monetary policy

The article provides a survey of the contemporary literature on the role of sentiment in finance, macroeconomics, and monetary policy. It discusses approaches to measuring sentiment, including the construction of sentiment indices based on the analysis of unstructured data of various types (text, audio, and video), and examines the use of sentiment indicators for forecasting macroeconomic variables. The paper demonstrates that the expanding application of sentiment in microeconomic, macroeconomic, and financial analysis is closely linked to recent advances in machine learning methods for processing unstructured data. The survey shows that sentiment is closely related to the mechanisms of expectation formation and can substantially amplify macroeconomic fluctuations. This effect may be explained by the bounded rationality of economic agents: excessive reactions to recent news may arise due to the behavioral representativeness heuristic. In the context of macroeconomic and financial fluctuations, such overreaction is described by the diagnostic expectations hypothesis, which appears to be broadly consistent with Russian data. Particular attention is devoted to models of inflation expectations and to empirical findings, including estimates for the Russian economy, that support the view that the transmission of monetary policy changes in the presence of behavioral distortions. The paper concludes that sentiment and communication effects should be explicitly taken into account when forecasting the consequences of monetary policy.