The News Tone as a Leading Indicator of Consumer Sentiment

The article presents an approach to the use of text analysis in order to form quantitative leading indicators based on qualitative data. In the theoretical part of the article, the author provides an overview of leading papers by foreign researchers, including those from central banks of developed countries, devoted to the assessment and use of data on the tone of news publications as an indicator of economic activity. Various approaches to extracting the tone of texts and converting its qualitative assessments into quantitative indicators are covered. The author's approach to evaluating the news tone using machine learning methods is explained. In the empirical part of the article, the author uses the example of the consumer sector to test the hypothesis about the consistency of fluctuations in the emotional tone of news publications with fluctuations in key indicators of the state of the economy of Russia. A news indicator based on publications in online media devoted to the retail and services sector is generated. As a result of the study, the consistency of its dynamics with individual indicators of consumer sentiment, calculated on the basis of surveys conducted by "inFOM" LLC, is confirmed. The conclusions presented in the works of foreign researchers regarding the consistency of the dynamics of consumer sentiment, measured on the basis of sociological surveys, and synthetic indicators reflecting the emotional tone of the news background are validated.

text analysis, news sentiment, leading indicator, news tone, macroeconomics, macroeconomic indicators