Time-Series and Dynamic Cross-Correlations Analysis on Unexpected Information: Evidence from Media News and Online Postings

In this paper, we investigate the relationship between unexpected information from postings and news, and the unexpected information is measured by the residual of regressions of trading volume on numbers of news or postings. We mainly find that (i) There are significant positive contemporaneous correlations between the unexpected information coming from postings and different kinds of news; the correlation between the unexpected information coming from postings and new media news is stronger than that between the unexpected information coming from postings and mass media news; (ii) The unexpected information coming from postings could cause the unexpected information coming from news, but only the unexpected information coming from the mass media news could cause that coming from postings; (iii) There are persistent power-law cross-correlations between the unexpected information coming from postings and that coming from mass media news and new media news. The cross-correlation between the unexpected information coming from postings and new media news is more persistent than the one between the unexpected information coming from postings and mass media news. The cross-correlations are all more stable in long term than in short term. We attribute our findings above to the dissemination speed of the information on the Internet.