The cross-correlations between online sentiment proxies: Evidence from Google Trends and Twitter

In this paper, we explore the cross-correlations between two commonly-employed online sentiment proxies, i.e., the Financial and Economic Attitudes Revealed by Search (FEARS) from Google Trends and Daily Happiness Sentiment (DHS) from Twitter, with the methodology of MF-DCCA. The empirical results mainly show that: firstly, there exists power-law cross-correlation between the FEARS and DHS and the cross-correlation between them perform multifractality; secondly, the degree of multifractality in short term is significantly smaller than that in long term indicating a more stable cross-correlation in short term; finally, with the rolling window analysis, we further find that the evolution of the cross-correlations between FEARS and DHS is erratic.