ИСТИНА

This paper proposes a new approach to measure inflation expectations by the example of the Russian economy. It based on text mining of information on the Internet and machine learning techniques. We analyzed all news and articles about inflation and readers comments on them for major Russian economic media available in the web at the period from 2014 through 2016 and constructed two indicators reflecting inflation expectations dynamics. The first indicator was built on the base of text frequency analysis. It shows the intensity of inflation expectations in the short term (up to 1 year) for the Russian population. The second indicator was evaluated by means of sentiment analysis of comments content. This indicator in turn reflects the degree of inflation ex-pectations uncertainty. During the whole time period both indicators were highly correlated with key macroeconomic indicators, including official survey-based Bank of Russia inflation expecta-tions indicators. Moreover, new indicators are also able to forecast dynamics of official indicators for approximately 1 month in advance. Key advantages of our new approach in relation to current best practice methods – survey-based and market-based approaches – consist in: ability to update computations in real-time and to move ahead of survey-based indicators, wide coverage of the rel-evant audience (economic agents), relatively low cost of all computations and possibility to explore reasons of any changes in inflation expectations. By applying word frequency analysis and topic modeling algorithms to daily representation of new indicators, we were able to reveal key factors of inflation expectations dynamics during the period from 2014 till 2016. For each factor we were able to trace the changes in its relative contribution in inflation expectations over time. The final list of inflation expectations factors included the following categories: national exchange rate, commodity prices, monetary policy, economic situation uncertainty, economic crisis, domestic pol-icy and the world economic situation.