QUIET: an Indicator of Inflation Expectations of Households in Russia and Its Regions Based on Search Engine Data
Karpeko F.
This study proposes a novel indicator of households’ inflation expectations, constructed from search engine data for Russia and for its regions. First, we select several key topics of search queries related to inflation dynamics and expectations. Second, using principal component analysis, construct an aggregate indicator of inflation expectations at both monthly and weekly frequencies. Third, we examine the relationship between the presented indicator and the dynamics of the Consumer Price Index, comparing our findings with those obtained for the inFOM household inflation expectations and for the price expectations of firms participating in the Bank of Russia’s Monitoring of Businesses. Finally, we produce CPI forecasts using an expanding-window approach that incorporates the search-query-based indicator, in particular by employing weekly data within MIDAS models.
The new indicator exhibits a high degree of explanatory power in models with price dynamics as a dependent variable. Moreover, its inclusion in forecasting models reduces the forecast error over short-term horizons. In addition, the indicator can be updated almost in real time, making it suitable for nowcasting purposes. The results suggest that data from search engines, as a source of up-to-date information on household sentiment and expectations, can be a useful complement to inflation expectations measures obtained from sociological surveys.